UK e-Science Technical Report Series ISSN 1751-5971 GridNet2 Final Report EPSRC/JISC Principal Investigator: Prof Malcolm Atkinson Co-Investigator: Dr Anna Kenway Edited by Katharine Woods 31-Jul-2008 Abstract: GridNet2 was designed to fund travel and attendance for standards work in the international e-Science community, extending the UK's influence and seeking to ensure the needs of the UK's e-Infrastructure were incorporated into the developing standards. GridNet2 members participated at all levels, as leaders of and contributors to the working, research and community groups in the Standards Development Organisations (SDOs), in the international e-Science and Grid communities, driving standards development forward and promoting this work in the academic, public and industrial sectors. As a condition of receiving funding, GridNet2 members were required to write reports on the SDO meetings and conferences that they were funded to attend. This report collates all the individual reports and contains summary data for each GridNet2 member based on those reports. It also contains summary data for GridNet2 as a whole, including lists of publications, participation in standards work by area, events, and financial awards by individual and institution. The report also reviews the management and administration of GridNet2. UK e-Science Technical Report Series Report UKeS-2008-04 Available from http://www.nesc.ac.uk/technical_papers/UKeS-2008-04.pdf Copyright © 2008 The University of Edinburgh. All rights reserved. Funded By: Start Date: End Date: EPSRC (EP/D022169/1) and JISC 1st August 2005 30th April 2008 Contents Contents 1 GridNet2 Summary.......................................................................... 10 1.1 GridNet2 Background ................................................................... 10 1.2 GridNet2 Aims............................................................................. 10 1.3 GridNet2 Administration ............................................................... 11 1.3.1 GridNet2 Investigators........................................................... 11 1.3.2 GridNet2 Advisory Board ........................................................ 12 1.3.3 GridNet2 Management ........................................................... 13 1.4 2 GridNet2 Achievements ................................................................ 14 GridNet2 Member Summaries ......................................................... 19 2.1 Mr Oluwafemi Ajayi – NeSC .......................................................... 20 2.2 Dr Mario Antonioletti – EPCC......................................................... 20 2.3 Prof Malcolm Atkinson – NeSC ...................................................... 21 2.4 Prof Mark Baker – Reading ........................................................... 23 2.5 Dr Dave Berry – NeSC ................................................................. 23 2.6 Prof David Chadwick – Kent.......................................................... 24 2.7 Mr Xiaoyu Chen – Brunel.............................................................. 26 2.8 Mr Andrew Cooper – Oxford ......................................................... 26 2.9 Dr Thierry Delaitre – Westminster ................................................. 27 2.10 Dr Abdeslem Djaoui – STFC ...................................................... 27 2.11 Mr Donal Fellows – Manchester .................................................. 28 2.12 Dr Steve Fisher – STFC............................................................. 29 2.13 Mr Andrew Harrison - Cardiff ..................................................... 30 2.14 Mr Chris Higgins – Edinburgh .................................................... 30 2.15 Mr Neil Chue Hong – OMII-UK ................................................... 31 2.16 Dr Richard Hughes-Jones – Manchester ...................................... 32 2.17 Dr Jens Jensen – STFC ............................................................. 33 2.18 Dr Shantenu Jha – UCL............................................................. 34 2.19 Mr Jipu Jiang – NeSC................................................................ 35 2.20 Dr Mike Jones – Manchester ...................................................... 35 2.21 Mr Ian Kelley – Cardiff.............................................................. 36 2.22 Dr Amrey Krause – Edinburgh ................................................... 37 2.23 Mr William Lee – Imperial ......................................................... 38 2.24 Mr Mark Leese – STFC .............................................................. 39 2.25 Dr Stephen McGough – Imperial ................................................ 40 2.26 Dr Andrew Martin – Oxford ....................................................... 41 Page 3 Contents 2.27 Dr Steven Newhouse – OMII-UK ................................................ 42 2.28 Mr Vesselin Novov – Imperial .................................................... 43 2.29 Dr Savas Parastatidis - Newcastle .............................................. 44 2.30 Dr Colin Perkins – Glasgow ....................................................... 45 2.31 Dr Stephen Pickles – Manchester ............................................... 46 2.32 Prof Omer Rana – Cardiff .......................................................... 48 2.33 Dr Matthew Shields – Cardiff ..................................................... 49 2.34 Prof Richard Sinnott - NeSC ...................................................... 50 2.35 Mr David Spence – STFC........................................................... 51 2.36 Mr Anthony Stell – NeSC........................................................... 52 2.37 Dr Ian Taylor – Cardiff.............................................................. 53 2.38 Mr Elias Theocharopoulos – NeSC............................................... 53 2.39 Dr David Wallom – Oxford ........................................................ 54 2.40 Dr Viktor Yarmolenko – Manchester............................................ 55 2.41 Mr Stefan Zasada – UCL ........................................................... 56 2.42 Dr Ning Zhang – Manchester ..................................................... 56 3 GridNet2 Reports ............................................................................ 58 3.1 Mr Oluwafemi Ajayi – NeSC .......................................................... 58 3.1.1 GGF16 ................................................................................ 58 3.1.2 GGF18 ................................................................................ 59 3.1.3 OGF20 ................................................................................ 60 3.2 Dr Mario Antonioletti - EPCC ......................................................... 60 3.2.1 3.3 Prof Malcolm Atkinson – NeSC ...................................................... 62 3.3.1 GFSG F2F............................................................................ 62 3.3.2 GGF16 and GFSG ................................................................. 62 3.3.3 GGF17 ................................................................................ 63 3.3.4 GGF 18 ............................................................................... 64 3.3.5 OGF19 ................................................................................ 65 3.3.6 OGF20 ................................................................................ 66 3.3.7 OGF21 ................................................................................ 66 3.4 Prof Mark Baker – Reading ........................................................... 67 3.4.1 3.5 Page 4 OGF22 ................................................................................ 60 ICEAGE - Curricula Development Workshop.............................. 67 Dr Dave Berry – NeSC ................................................................. 67 3.5.1 OGSA F2F – Jan 06............................................................... 67 3.5.2 GGF16 ................................................................................ 67 3.5.3 GGF18 ................................................................................ 68 3.5.4 OGSA WG Telcons – Aug 06 ................................................... 68 Contents 3.5.5 3.6 OGF19 ................................................................................ 68 Prof David Chadwick – Kent.......................................................... 68 3.6.1 GGF14 ................................................................................ 68 3.6.2 GGF15 ................................................................................ 69 3.6.3 GGF16 ................................................................................ 69 3.6.4 GGF17 ................................................................................ 70 3.6.5 GGF18 ................................................................................ 71 3.6.6 OGF19 ................................................................................ 71 3.6.7 OGF20 ................................................................................ 73 3.6.8 OGF21 ................................................................................ 73 3.6.9 OGF22 ................................................................................ 75 3.7 Mr Xiaoyu Chen – Brunel.............................................................. 75 3.7.1 OGF21 ................................................................................ 75 3.8 Mr Andrew Cooper – OERC ........................................................... 76 3.9 Dr Thierry Delaitre – Westminster ................................................. 76 3.9.1 GGF18 ................................................................................ 76 3.10 Dr Abdeslem Djaoui – STFC ...................................................... 77 3.11 Mr Donal Fellows – Manchester .................................................. 77 3.11.1 3.12 OGF21 and OGF22 ............................................................ 77 Dr Steve Fisher – STFC............................................................. 77 3.12.1 GGF15............................................................................. 77 3.12.2 GGF16............................................................................. 80 3.12.3 GGF17............................................................................. 80 3.12.4 INFOD-WG F2F – May 06 ................................................... 82 3.12.5 GGF18............................................................................. 82 3.12.6 INFOD-WG F2F – February 07............................................. 83 3.12.7 OGF19............................................................................. 83 3.12.8 OGF20............................................................................. 83 3.12.9 OGF21............................................................................. 83 3.13 Mr Andrew Harrison – Cardiff..................................................... 84 3.13.1 OGF20............................................................................. 84 3.13.2 OGF21............................................................................. 85 3.13.3 OGF22............................................................................. 86 3.14 Mr Chris Higgins – Edinburgh .................................................... 87 3.14.1 OGC Committee Meeting – June 06...................................... 87 3.14.2 GGF18............................................................................. 87 3.14.3 OGF20............................................................................. 88 3.14.4 OGF21............................................................................. 89 Page 5 Contents 3.14.5 3.15 Mr Neil Chue Hong – OMII-UK ................................................... 91 3.15.1 OGF21............................................................................. 91 3.15.2 OGF22............................................................................. 92 3.16 Dr Richard Hughes-Jones – Manchester ...................................... 92 3.16.1 GGF15............................................................................. 92 3.16.2 OGF19............................................................................. 94 3.16.3 OGF20............................................................................. 94 3.16.4 OGF21............................................................................. 95 3.17 Dr Jens Jensen – STFC ............................................................. 96 3.17.1 3.18 OGF22............................................................................. 96 Dr Shantenu Jha – UCL............................................................. 99 3.18.1 GGF15............................................................................. 99 3.18.2 GGF16........................................................................... 100 3.18.3 GGF17........................................................................... 100 3.18.4 GGF18........................................................................... 100 3.18.5 LGC Workshop ................................................................ 100 3.18.6 SAGA Design and Implementation Meeting ......................... 101 3.18.7 ISSGC 07....................................................................... 101 3.19 Mr Jipu Jiang – NeSC.............................................................. 101 3.19.1 3.20 GGF18........................................................................... 101 Dr Mike Jones – Manchester .................................................... 102 3.20.1 GGF16........................................................................... 102 3.20.2 GGF17........................................................................... 102 3.20.3 GGF18........................................................................... 102 3.20.4 OGF20........................................................................... 103 3.20.5 OGF21........................................................................... 103 3.20.6 OGF22........................................................................... 104 3.21 Mr Ian Kelley – Cardiff............................................................ 105 3.21.1 GGF18........................................................................... 106 3.21.2 OGF20........................................................................... 106 3.21.3 OGF21........................................................................... 106 3.21.4 OGF22........................................................................... 107 3.22 Page 6 OGF22............................................................................. 89 Dr Amrey Krause – Edinburgh ................................................. 107 3.22.1 GGF16........................................................................... 107 3.22.2 GGF18........................................................................... 108 3.22.3 OGF20........................................................................... 108 3.22.4 OGF21........................................................................... 109 Contents 3.22.5 OGF22........................................................................... 110 3.23 Mr William Lee - Imperial........................................................ 110 3.24 Mr Mark Leese – STFC ............................................................ 110 3.25 Dr Stephen McGough – Imperial .............................................. 111 3.25.1 GGF14........................................................................... 111 3.25.2 GGF15 & OGSA F2F ......................................................... 111 3.25.3 GGF16 & OGSA F2F ......................................................... 112 3.25.4 GGF17........................................................................... 112 3.25.5 GGF18........................................................................... 113 3.25.6 OGF19........................................................................... 114 3.25.7 OGF20........................................................................... 115 3.25.8 OGF21........................................................................... 118 3.25.9 OGF22........................................................................... 120 3.25.10 OGSA F2F (12) ............................................................... 123 3.25.11 OGSA F2F (13) ............................................................... 123 3.25.12 OGSA F2F (14) ............................................................... 123 3.25.13 OGSA F2F (15) ............................................................... 123 3.25.14 OGSA F2F (17) ............................................................... 124 3.25.15 OGSA F2F (18) ............................................................... 125 3.25.16 OGSA F2F (19) ............................................................... 126 3.25.17 SuperComputing ‘06 and OGSA F2F (16) ............................ 127 3.25.18 SuperComputing ‘07 ........................................................ 127 3.26 Dr Andrew Martin – Oxford ..................................................... 128 3.26.1 GGF15........................................................................... 128 3.26.2 GGF16........................................................................... 129 3.26.3 Trusted Computing Group Meeting..................................... 130 3.27 Dr Steven Newhouse – OMII-UK .............................................. 130 3.27.1 GFSG F2F – Jan 06.......................................................... 130 3.27.2 OGSA F2F – Jan 06 ......................................................... 130 3.27.3 GGF16........................................................................... 131 3.27.4 GGF17........................................................................... 131 3.28 Mr Vesselin Novov – Imperial .................................................. 132 3.29 Dr Savas Parastatidis – Newcastle............................................ 132 3.29.1 3.30 GGF14........................................................................... 132 Dr Colin Perkins – Glasgow ..................................................... 132 3.30.1 IETF64 .......................................................................... 132 3.30.2 IETF65 .......................................................................... 133 3.30.3 IETF66 .......................................................................... 134 Page 7 Contents 3.30.4 IETF67 .......................................................................... 134 3.30.5 IETF68 .......................................................................... 134 3.30.6 IETF69 .......................................................................... 135 3.30.7 IETF70 .......................................................................... 135 3.30.8 SUMOVER – Nov 05......................................................... 136 3.30.9 SUMOVER – Apr 07 ......................................................... 136 3.31 3.31.1 OGF .............................................................................. 137 3.31.2 OASIS ........................................................................... 137 3.31.3 ETSI.............................................................................. 137 3.31.4 GGF15........................................................................... 137 3.31.5 GGF16........................................................................... 137 3.31.6 GGF17........................................................................... 137 3.31.7 GGF18........................................................................... 137 3.31.8 OGF19........................................................................... 137 3.31.9 OGF20........................................................................... 137 3.31.10 GFSG F2F – Jan 06.......................................................... 137 3.31.11 GFSG F2F – Jul 06........................................................... 138 3.31.12 GFSG F2F – Dec 06 ......................................................... 138 3.31.13 OGF GIN F2F – Mar 07 ..................................................... 138 3.31.14 ETSI Meeting – Nov 06 .................................................... 138 3.32 Prof Omer Rana – Cardiff ........................................................ 138 3.32.1 GGF16........................................................................... 138 3.32.2 GGF17........................................................................... 141 3.32.3 GGF18........................................................................... 145 3.32.4 OGF19........................................................................... 148 3.32.5 OGF20........................................................................... 150 3.32.6 OGF21........................................................................... 152 3.32.7 OGF22........................................................................... 152 3.33 Dr Matthew Shields – Cardiff ................................................... 154 3.33.1 OGF20........................................................................... 154 3.33.2 OGF21........................................................................... 155 3.34 Page 8 Dr Stephen Pickles – Manchester ............................................. 136 Prof Richard Sinnott – NeSC .................................................... 156 3.34.1 GGF15........................................................................... 156 3.34.2 GGF16........................................................................... 157 3.34.3 GGF18........................................................................... 158 3.34.4 OGF20........................................................................... 159 3.34.5 OGF21........................................................................... 160 Contents 3.34.6 3.35 Mr David Spence – STFC......................................................... 162 3.35.1 3.36 Dr Ian Taylor – Cardiff............................................................ 164 GGF18........................................................................... 164 3.37.2 OGF20........................................................................... 164 3.37.3 OGF21........................................................................... 165 3.37.4 OGF22........................................................................... 165 Mr Elias Theocharopoulos - NeSC ............................................. 166 3.38.1 3.39 OGF22........................................................................... 166 Dr David Wallom – OeRC ........................................................ 167 3.39.1 GGF18........................................................................... 167 3.39.2 OGF20........................................................................... 168 3.39.3 OGF22........................................................................... 169 3.40 Dr Viktor Yarmolenko – Manchester.......................................... 172 3.40.1 3.41 3.42 SLA F2F ......................................................................... 172 Mr Stefan Zasada – UCL ......................................................... 174 3.41.1 OGF20........................................................................... 174 Dr Ning Zhang – Manchester ................................................... 175 3.42.1 OGF19........................................................................... 175 3.42.2 OGF20........................................................................... 175 GridNet2 Financial Statements...................................................... 176 A.1 B GGF16........................................................................... 163 3.37.1 3.38 A GGF18........................................................................... 162 Mr Anthony Stell – NeSC......................................................... 163 3.36.1 3.37 OGF22........................................................................... 161 Financial Awards and Claims....................................................... 176 Publications.................................................................................. 179 B.1 Publications with GridNet2 Authors .............................................. 179 B.2 Publications with GridNet2 Contributors........................................ 181 C Standardisation Participation by Area .......................................... 184 D Events .......................................................................................... 191 D.1 GridNet2 e-Science Workshop (OGF21)......................................... 193 Glossary ............................................................................................... 195 Page 9 1 GridNet2 Summary 1 GridNet2 Summary This section provides a summary of the background, aims and achievements of the GridNet2 project, which ran from 1st August 2005 to 30th April 2008. 1.1 GridNet2 Background The GridNet2 project followed on from the GridNet1 project. GridNet1 was funded by EPSRC grant GR/R74772/01 and ran from February 2002 to February 2005. GridNet1 allowed the UK to establish a significant influence on the development of standards and to share best practices in e-Science standardisation. This engagement was invaluable as the future of e-Research and the generalisation of e-Science demanded a cost-effective, sustainable, persistent and reliable e-infrastructure. GridNet1 enabled the following UK contributions to the developing standards in e-Science: • UK staff and activities were established in leading roles in standards bodies. • UK practitioners of e-Sciences have influenced emerging standards. • UK researchers and developers were established within the international e-Science community, ensuring rapid communication of ideas, best practice and plans, which were important steps in making internationally sustained e-Infrastructure available to the UK research community. GridNet1 established a UK presence at the Global Grid Forum (GGF) whereby: • 46 people were registered and active at various GGF meetings. • One representative on the GFAC. • Five representatives on the GFSG. • UK leadership in 16 GGF working and research groups. • UK researchers had authored 9 of the 41 GGF documents published at the time of GridNet1 ending, with many further documents underway. Other GridNet1 achievements include community building, requirements analysis, collaborative demonstrations, workshops, informational document preparation, and standards specification proposals. 1.2 GridNet2 Aims As a result of the GridNet1 achievements, the UK was in a position where it had a greater potential influence on the GGF standards than any country other than the USA at the start of GridNet2. The aim of GridNet2 was to build on this potential by continuing to fund attendance at SDO meetings so that standardisation work begun during GridNet1 could continue in GridNet2. It was recognised that additional standardisation work was taking place outside the usual SDO meetings, and GridNet2 needed to support these meetings. Three examples of such meetings are as follows: • Page 10 Regular teleconferences (telcons) to progress standards and documents. 1 GridNet2 Summary For example, the OGSA WG of the GGF averaged 3 telcons per week for its subsidiary design teams, and the DAIS documents required telcons between the authors, editors, reviewers and contributors. • Face-to-face (F2F) meetings, taking place over a period of 2-4 days, allow participants to focus exclusively on the issues in developing a standard. Such meetings require hosting or travel. For examples of such meetings funded by GridNet2, see the following sections: • • Section 3, GridNet2 Reports • Appendix D, Events Document preparation meetings where sub-groups of editors and authors meet to consolidate the work that has been undertaken through mail lists, web forums and F2F meetings and telcons. GridNet2 also needed to support socialisation tours. Socialisation tours are used to enhance the prospects for a standard to be widely adopted by visiting relevant projects, middleware development teams and industry decision makers. Where standards have become of greater importance to industry and commerce, GridNet2 funding was required, in UK interests, to keep in touch with the progress, evaluation and assessments of these standards and their rivals. UK researchers and developers had obtained senior positions in various SDOs, and GridNet2 funding was required to co-ordinate and support these efforts so that UK investment in the SDOs had increased effect. The intention was also to avoid extra time commitments for people already heavily involved in standards, and reducing enthusiasm where therewas already effective engagement, such as with JSDL and SAGA. 1.3 GridNet2 Administration 1.3.1 GridNet2 Investigators Prof. Malcolm Atkinson, Director of the National e-Science Centre, e-Science Institute, University of Edinburgh, was the Principal Investigator (PI) for GridNet2. Prof Atkinson took overall financial and managerial responsibility for GridNet2, and was responsible for the following aspects of GridNet2: • Providing direction and leadership • Co-ordinating the work of the Advisory Board for GridNet2 • Advising on applications • Communicating with the UK e-Science Core Programme. Dr Anna Kenway, Deputy Director of the National e-Science Centre, e-Science Institute, University of Edinburgh, was the Co-Investigator for GridNet2. Dr Kenway was responsible for the following aspects of GridNet2: • Appointing and supervising staff engaged on GridNet2 • Directing and overseeing operations, such as: • • Financial book-keeping • Web presentation Communicating with applicants and institutions Page 11 1 GridNet2 Summary 1.3.2 GridNet2 Advisory Board A GridNet2 Advisory Board (GNAB2) was set up to monitor and review the allocation process. The terms of reference for the GNAB2 included the membership and roles. The GNAB2 membership comprised the following: • The director of the UK e-Science Programme or his nominee • Eight members selected from or nominated by the directors of the UK e-Science Centres, Centres of Excellence and CCLRC • The directors of the NERC Environmental e-Science Institute, the Open Middleware Infrastructure Institute (OMII-UK), the Digital Curation Centre (DCC), National Grid Service (NGS) and the ESRC National Centre for e-Social Science (NCeSS) or their nominees • Five of the UK staff actively engaged in UK standards efforts • The PI of the GridNet2 project Table 1: Page 12 GridNet2 Advisory Board Members Name Institution Prof Malcolm Atkinson Director National e-Science Centre, U. Edinburgh Prof David Chadwick Deputy Director ESNW, U. Manchester Mr Neil Chue Hong Director, OMII-UK (replacement for Dr Steven Newhouse) Prof Jon Crowcroft U. Cambridge Computing Lab Dr Martin Dove Director, NIEeS, Cambridge Mr Alex Hardisty Manager, Welsh e-Science Centre Prof Peter Henderson PI OMII Dr Shantenu Jha Research Fellow, UCL Prof Andy Keane Director, Southampton e-Science Institute Dr Stephen McGough Technical Co-ordinator, LeSC, Imperial College Dr Steven Newhouse Director, OMII-UK (replaced by Mr Neil Chue Hong) Dr Savas Parastatidis NEReSC and OMII Prof Andy Parker Director, Cambridge e-Science Centre Prof Ron Perrott Director, Belfast e-Science Centre Prof Rob Procter Director, NCeSS, U. Manchester Dr Chris Rusbridge Director, DCC, U. Edinburgh Dr Richard Sinnott Deputy Director, Glasgow NeSC Dr Arthur Trew Director, EPCC, U. Edinburgh 1 GridNet2 Summary The role of the GNAB2 members was: 1. To review applications for GridNet2 funding, as in GridNet1. There were three anonymous reviews for each application, and reviewers could choose one of the following responses: • Fund in full • Fund to a specified level • Don’t fund. It was sometimes necessary to seek clarifications from applicants. This was undertaken via the NeSC GridNet2 administrator. If the reviewers did not reach agreement, a further reviewer from the GNAB2 was consulted at the discretion of the PI. The requests for reviews were always sent to institutions not involved in the application. 2. To advise on policy for the allocation of GridNet2 funds and strategic standards issues, either via the quarterly GridNet2 Access Grid (AG) meetings or via e-mail with the GridNet2 PI. These arrangements were reviewed and revised by the GNAB2 from time-totime, and changes were announced on the GridNet2 web pages (http://www.nesc.ac.uk/nesc/gridnet/). The GridNet2 application, review and reporting process was supported by web forms (for example, see http://www.nesc.ac.uk/action/funding/fa_forma.cfm?id=2) and a database for the applications, comments and decisions. The GridNet2 Advisory Board held meetings in: 1.3.3 • June 2006 • September 2006 • January 2007 • May 2007 • January 2008 • May 2008 GridNet2 Management In order to build on the success of GridNet1, existing activities from GridNet1 were continued in GridNet2 and new ones were added for GridNet2. The following activities were continuations from GridNet1: • Appointment of the GNAB2, with a larger membership than that of GridNet1, in order to spread the review load. • Wide advertising of the available GridNet2 resources to the UK community, through the NeSC website and newsletter, directors meetings, All Hands Meetings (AHMs), and where other opportunities arose. • Continuation of the application, budget allocation and monitoring process, including web site maintenance, database infrastructure, bookkeeping, communication and process oversight. The following activities were new in GridNet2: • Tabulation of all known UK participants in relevant standards processes and their roles Page 13 1 GridNet2 Summary • Reports from all UK participants for each meeting they have attended. For more information, see section 3, GridNet2 Reports. • UK Standardisation AG meetings to co-ordinate UK standards work • Specialist standardisation e-mail lists, web forums, AG meetings, telcons, and F2F meetings where the need arose to develop UK strategies, consensus or adoption of specific issues or standards • Support for socialisation of UK standards • Support for hosting (and in exceptional circumstances, attending) regular international telcons GridNet2 management also included the following management activities: 1.4 • Managing the sponsorship and membership of SDOs • Supporting SDO steering group members • Fast-tracking funding requests for funding amounts up to £1000, where an immediate response is required GridNet2 Achievements Table 2 gives summary data for GridNet2. Detailed breakdowns of the information in this table are given in later sections of this report. Table 2: GridNet2 Summary Data Number of Awards No. Further Information 45 Some GridNet2 members had more than one award. See section 2, GridNet2 Member Summaries Number of People Funded 42 Some GridNet2 awards funded more than one person. See section 2, GridNet2 Member Summaries. Events Supported 44 See appendix D, Events. Roles 249 See section 2, GridNet2 Member Summaries. Publications 59 See appendix B, Publications. Groups 49 See appendix C, Standardisation Participation by Area. Number of Claims 42 There were three GridNet2 awards for which no claims were submitted. During the course of GridNet2, members (those who applied for and received GridNet2 funding) participated in meetings and conferences, both F2F and via telcons, and engaged in SDO research, community and working groups to continue the work in developing e-Science standards. Standards have been published, revised and extended in this period, and GridNet2 members have undertaken socialisation work to disseminate awareness and increase uptake of these standards in academic, research and industrial areas. Page 14 1 GridNet2 Summary Standards that have been a particular success during GridNet2 are the Job Submission Description Language (JSDL) and A Simple API for Grid Application (SAGA) standards. Many GridNet2 members have participated in this work, with Dr Stephen McGough being extensively involved in JSDL and Dr Shantenu Jha in SAGA. JSDL v1.0 was published in November 2005. The work of the JSDL continued throughout GridNet2, and the publication of JSDL v1.0 was followed by HPC Profile Application Extension for JSDL 1.0 and the SPMD Application Extension for JSDL v1.0, both published in August 2007. Dr Stephen McGough is one of the authors for JSDL v1.0, and an acknowledged contributor for the SPMD Application Extension, along with Mr Donal Fellows, Mr William Lee and Dr Stephen Pickles. Mr Donal Fellows is also an acknowledged contributor for the HPC Profile Extension Application. In all, seven GridNet2 members participated in the JSDL work. JSDL has had considerable success, as indicated by the fact that there are more than twelve known implementations of JSDL v1.0. These implementations, created on a variety of platforms, have been demonstrated at the SuperComputing conferences held in 2006 and 2007. There were also ten implementations of the HPC Profile Extension to JSDL successfully demonstrated at SuperComputing 2006 and and 2007. Following on from JSDL v1.0, the JSDL-WG actively sought and received feedback on JSDL which led to the published extensions and errata for JSDL v1.0. In acknowledgement of the effort behind and success of JSDL, the JSDL-WG was presented with the OGF Leadership Award at OGF20. The award is given to individuals or groups within OGF that have demonstrated outstanding leadership and many contributions to the OGF mission. This was a ringing endorsement from the Grid community of the significance and value of the completed and ongoing work of the JSDL-WG. The SAGA-RG worked on the SAGA standard throughout the GridNet2 period, with participation from 13 GridNet2 members, and the SAGA standard was published in January 2008. Dr Shantenu Jha was one of the authors for this standard, and Dr Steven Newhouse and Dr Stephen Pickles were acknowledged contributors. This publication of the SAGA standard followed SAGA outreach that took place at ISSGC07, which has led to an investigation of use of SAGA as an interface for Condor. The work of the NM-WG has also been a success, as indicated by the 29 networks worldwide that have adopted work of the group. Dr Richard Hughes-Jones and Mr Mark Leese have been active leaders in this area and are co-chairs of the NM-WG. The OGSA-DAI standard has also been adopted by many groups, particularly in Asia. Various GridNet2 members have been involved in the OGSA-DAI project, including Prof Malcolm Atkinson, Dr Marion Antonioletti, MrNeil Chue Hong, Dr Amrey Krause, Dr Savas Parastatidis, Prof Richard Sinnott and Mr Elias Theocharopoulos. The OGSA-DAI project has close involvement with the DAIS-WG at OGF, and Dr Mario Antonioletti is one of the co-chairs of DAIS-WG. Dr Dave Berry, Dr Savas Parastatidis, Mr Neil Chue Hong, Dr Amrey Krause and Mr Elias Theocharopoulos have also participated in the DAIS-WG sessions at GGF/OGF meetings. Page 15 1 GridNet2 Summary It is not only through OGF that GridNet2 members have participated through the development of standards that benefit UK e-Science and the increased standing of UK e-Science in the global Grid and e-Science communities. Dr Colin Perkins has taken an active and leading role within the IETF, cochairing both the AVT and MMUSIC working groups. Dr Perkins has also developed a new proposal for the transport of real-time audio/visual data using RTP over the DCCP standard. Dr Perkins has also undertaken socialisation work for this standard and the SDP specification at SUMOVER workshop, prior to the adoption of these standards by the IETF. GridNet2 members have participated actively in the area of published standards, leading and directing the development of these standards. This leadership role has not only been in the area of published standards, but also in sustaining the ongoing work of the standards bodies. GridNet2 members have driven the establishment of new areas and groups within the e-Science standards community, consolidating and extending the presence of UK e-Science within the global community. During the course of GridNet2, David Wallom, one of the recipients of GridNet2 funding, was an active participant in setting up OGF-Europe. The setting up of OGF-Europe is another significant effort that will highlight the requirements and work of UK e-Science. The goal of OGF-Europe is to stimulate, co-ordinate and harmonise efforts to increase Grid adoption across Europe, and globally, through reinforcement of the Open Standards message. This will be achieved by mobilising the Grid European community of researchers, developers, providers, and end-users in both the public & private sectors. OGF-Europe aims to focus on issues that are of prime importance within the EU, ultimately increasing the ability of industry and commerce to influence requirements for a more competitive ICT infrastructure, built on these solid foundations. OGF-Europe will continue to shape the broad vision while supporting practical progress on maturing and standardising the technology based on real-world experience. The OGF-Europe consortium has a wide and varied membership from across Europe, but all have significant expertise, knowledge and skills to demonstrate how Grid infrastructures are a competitive advantage in Europe. The consortium also has significant cross linkage to the many aspects of the management structure of the OGF-Global organisation. Major outputs of OGF-Europe include: • 3 reports supporting the OGF technical & strategic roadmap • 2 survey reports on trends & Grid practices • 4 workshops on challenges on standardisation issues • 4 Grid adoption challenge reports • 6 Community outreach seminars including 4 in-depth tutorials • 6 Community best practice reports • 2 OGF-EUROPE international events (European-based OGF events) For more information about OGF-Europe, see: http://www.ogfeurope.eu/ GridNet2 members have also been instrumental in chartering new groups within OGF. Page 16 1 GridNet2 Summary Professor Malcolm Atkinson had a central and leading role in organising the BoFs concentrating on Grid education at two OGF meetings. These led directly on to the establishment of the ET-CG which has the remit to bring together practitioners in Grid-related education and training (E&T) to share and develop best practice, to stimulate greater investment in Grid-related E&T and above all, to build a mutually supportive community of Grid trainers and educators. Prof Malcolm Atkinson is one of the co-chairs of the ET-CG and has continued to promote and drive the ET-CG agenda at OGF, as well as working in the area of Grid education in other forums, such as the ICEAGE project. Another area in which GridNet2 members have taken a leading role is in the LoA BOFs which led to the establishment of the LoA-RG. Dr Ning Zhang took on the role of co-chair of the LOA-RG, after organising and chairing the LoA BoF at OGF19. Dr Mike Jones has also participated in the LoA-RG. The aim of the LoA-RG is to investigate use case scenarios in the e-Science/Grid contexts, and identify gaps in applying existing LoA definitions to such contexts. Table 3 lists the SDO working, community and research groups that GridNet2 members have chaired. There were 19 GridNet2 members who chaired or cochaired 20 working, research or community groups. Three GridNet2 members chaired or co-chaired two working groups. Table 3: GridNet2 Group Chairs Group Chair/Co-chair APP-AGG-WG Dr Ian Taylor AVT-WG Dr Colin Perkins BYTEIO-WG Mr Neil Chue Hong DAIS-WG Dr Mario Antonioletti ET-CG Prof Malcolm Atkinson GSM-WG Dr Jens Jensen INFOD-WG Dr Steven Fisher JSDL-WG Dr Stephen McGough LoA-RG Dr Ning Zhang MMUSIC-WG Dr Colin Perkins NM-WG Dr Richard Hughes-Jones Mr Mark Leese OGSA-Authz-WG Prof David Chadwick OGSA-BES-WG Dr Steven Newhouse OGSA-D-WG Dr Dave Berry OGSA-DMI-WG Dr Mario Antonioletti PGS-RG Dr David Wallom Page 17 1 GridNet2 Summary Group Chair/Co-chair SAGA-RG Dr Shantenu Jha TC-RG Dr Andrew Martin UR-WG Mr Xiaoyu Chen Mr Donal Fellows WFM-RG Dr Ian Taylor GridNet2 members have also taken other significant leadership roles in the standards organisations for the Grid community. Prof Malcolm Atkinson served on the OGF’s GFSG as Data Area Director, and then on the OGF Board of Directors. Dr Richard Hughes-Jones, Dr Steven Newhouse and Dr Stephen Pickles also served as OGF Area Directors for the Infrastructure, Applications and Compute Areas respectively. Mr Neil Chue Hong is also chair of the OGF Nominations Committee, which solicits and selects or recommends candidates to fill OGF leadership positions. OGF20 took place in Manchester, hosted by UK e-Science. GridNet2 members were instrumental in organising this OGF meeting, which had a very high attendance (over 900 attendees, compared with over 600 for GGF18 and over 250 for OGF21). Dr Stephen Pickles was chair of the Local Organising Committee for OGF20, and the co-located 2nd EGEE User Forum. Holding an OGF meeting in the UK gave an opportunity to showcase UK and European e-Science to the global Grid community. GridNet2 members (Prof Omer Rana, Dr Stephen McGough and Dr Ian Taylor) organised a GridNet2 e-Science workshop at OGF21. The aim of the workshop was to disseminate information about the work being done by GridNet2 members to both other GridNet2 members and the broader Grid community, and to encourage communication between the different areas of OGF in which GridNet2 members participate. The GridNet2 e-Science workshop was held over three sessions at OGF21, and in total, 19 people attended the three sessions. Each of the three sessions had a theme: • Use of novel computer science techniques • Demonstration provided through applications of e-Science concepts • Standards activities and coordination across standards bodies The workshop presentations are available from the wiki set up by Prof Omer Rana. For more information, see appendix D.1, GridNet2 e-Science Workshop (OGF21). Page 18 2 GridNet2 Member Summaries 2 GridNet2 Member Summaries This section contains the summaries for those who applied for and received GridNet2 funding (GridNet2 members). The information in this section is derived from the reports in section 3, GridNet2 Reports. There is a table for each GridNet2 member which includes the following information: • The People row shows the number of people, and their names and institution that were covered by the award. Where an award covers more than one person, there is a summary table for each person. • The Events entry lists the events that this GridNet2 member attended as a result of the GridNet2 award. For a list of events covered by GridNet2 funding, see appendix D, Events. • The Roles entry includes chairing or co-chairing meetings, chairing or co-chairing research, working and community groups, and giving presentations. The roles are intended to show the depth of participation in a particular area during GridNet2. The totals for roles have been calculated by each role per meeting. For example, if a GridNet2 member has chaired a working group over the course of two meetings, this is counted as two roles. • The Publications entry includes documents and draft documents where the GridNet2 member was an author or contributor to OGF documents, and any other publications and papers where GridNet2 funding was instrumental in producing the documents. For full details of the publications, see appendix B, Publications. • The Areas of interest entry lists the areas in which the GridNet2 member has participated. The areas of interest are intended to show the breadth of participation during GridNet2. This includes research, working and community group sessions, BoF sessions and workshops at GGF/OGF meetings, IETF meetings and F2F meetings. The totals for areas of interest have been calculated once for all the meetings which the GridNet2 member attended. For example, if the GridNet2 member has participated in the same working group sessions at three meetings, this is counted as one area of interest. Areas of interest do not include talks, such as keynotes or plenaries. • The Claim entry shows the claim number given for GridNet2 administration purposes, the amount of the award and the amount claimed. For a list of technical areas covered by GridNet2 funding, see appendix C, Standardisation Participation by Area. Page 19 2 GridNet2 Member Summaries 2.1 Mr Oluwafemi Ajayi – NeSC Mr Ajayi was supported by Prof Richard Sinnott’s GridNet2 awards (97 and117). Table 4: GridNet2 Summary for Oluwafemi Ajayi No. Description People 1 Mr Oluwafemi Ajayi, NeSC, University of Glasgow Events 3 GGF16, GGF18 1, OGF20 Roles 1 Presenter - OGSA-DAI User Forum (OGF20) Publications 0 Areas of interest 11 CA-OPS WG (GGF16) Dynamic Level Service Agreement (OGF20) Enterprise Grid Requirements (OGF18) GRAAP WG (OGF18) Grid Interoperation (GGF16) Grids Mean Business (OGF20) LSG-WG (GGF16/GGF18) OGSA-Authz WG (GGF16) OGSA-DAI User Forum (OGF20) Shibboleth for Grids (OGF18) ShibGrid (GGF16) Claim 97 117 See section 2.34, Prof Richard Sinnott - NeSC. To see the full details of Mr Ajayi’s GridNet2 activities, see section 3.1, Mr Oluwafemi Ajayi – NeSC. 2.2 Dr Mario Antonioletti – EPCC Dr Mario Antonioletti was funded by Dr Steven Newhouse’s GridNet2 award (99). Table 5: GridNet2 Summary for Mario Antonioletti No. Description People 1 Dr Mario Antonioletti, Edinburgh Parallel Computing Centre Events 1 OGF22 1 At the time of GGF18, the merger between GGF and EGA had been completed and the new organisation named as the Open Grid Forum (OGF). However, this event was still branded as a GGF event, and this same terminology has been used in this document to refer to this event. Page 20 2 GridNet2 Member Summaries Roles No. Description 6 Chair – OGSA Data Architecture Future Directions Discussion (OGF22) Co-chair – OGSA-DAI WG Co-chair – DAIS WG (OGF22) Co-chair – OGSA-DMI (OGF22) Presenter – OGSA Data Architecture (OGF22) Presenter – DAIS Overviews (OGF22) Publications 0 Areas of interest 16 Cloud Systems BoF DAIS-WG Data Management Workshop DFDL-WG GIN Globus Software User Experiences Grid Usage and Productivity in HPC HPCP Specification Adoption OMII-Europe OGSA-DAI OGSA-DMI OGSA Data Architecture SAGA SAGA+DAIS SRM 2.2 Protocol Review The Encyclopedia of Life Claim 99 See section 2.27, Dr Steven Newhouse – OMII-UK. To see the full details of Dr Antonioletti’s GridNet2 activities, see section 3.2, Dr Mario Antonioletti - EPCC. 2.3 Prof Malcolm Atkinson – NeSC Table 6: GridNet2 Summary for Malcolm Atkinson People No. Description 2 • Prof Malcolm Atkinson, Director of the e-Science Institute, NeSC, University of Edinburgh • Dr Dave Berry, Deputy Director of the e-Science Institute, NeSC, University of Edinburgh See section 2.5, Dr Dave Berry – NeSC. Events 10 GFSG F2F – Jan 06, GGF16, GFSG at GGF16, GGF17, GFSG at GGF17, GGF18, GFSG at GGF18, OGF19, OGF20, OGF21 Page 21 2 GridNet2 Member Summaries Roles No. Description 19 Area Director for the Data Area (GGF16, GGF17) Co-chair – Development of National and International Education and Training Policy (OGF19, OGF20, OGF21) Co-chair ET-CG (GGF18, OGF19, OGF20, OGF21) Co-chair Data Area Meeting (GGF17, GGF18) Co-chair – IPR for Grid Education and Training (OGF19) Co-chair – Policies for Cooperation on t-Infrastructure (OGF19, OGF20) Co-organiser – Workshops and BoFs on Grid Education and Training (GGF16, GGF17, GGF18) GFSG member Presenter – Grid Education and Training: Work in Progress and Future Collaboration (GGF16) Publications 1 Policy for Supporting Grid Education and Training Areas of interest 33 BYTEIO-WG (GGF17, GGF18, OGF21) DAIS-WG (GGF16, GGF17, GGF18, OGF19) DAIS for RDF/Ontology (GGF17, GGF18) Data Access and Integration (GGF16, GGF17) Data Area meeting (GGF16, GGF17) DFDL-WG (GGF16, GGF18) E&T Content (OGF19) E&T Frameworks (OGF19, OGF21) Earth Observation and Ground Systems (GGF18) Education and Training BoFs/ET-CG (GGF16, GGF17, GGF18) GFS-WG (GGF16, GGF17, GGF18) GIN-CG (OGF21) Grid Architecture (GGF16) GRIDFTP-WG (GGF16, GGF18) GSM-WG (GGF16, GGF18) HPCP-WG (GGF18, OGF21) INFO-D-WG (GGF16, GGF17) IPR for Grid Education (OGF19, OGF20, OGF21) Metadata, Search and Preservation in Data Grids (GGF17) OGC-OGF (GGF18) OGSA-BES-WG (GGF18) OGSA-D-WG (GGF16) OGSA-DAI-WG (GGF17) OGSA-DAI Users Forum (GGF18) OGSA Naming Report-Out (OGF20) OMII Core Grid Infrastructure (OGF19) Policy for Cooperation on t-Infrastructure (OGF19, OGF20, OGF21) Page 22 2 GridNet2 Member Summaries No. Description Preservation Environments in Data Grids (GGF17) Reference Model (OGF20) Standards All-Hands: Integrating the Work of OGF (OGF19) SNIA (GGF16) Towards Professional Grid Qualifications (OGF20) Voice of Community (GGF17) 100 Claim Awarded: £30,000 Claimed: £31,623.01 To see the full details of Prof Atkinson’s GridNet2 activities, see section 3.3, Prof Malcolm Atkinson – NeSC. 2.4 Table 7: Prof Mark Baker – Reading GridNet2 Summary for Mark Baker No. Description People 1 Prof Mark Baker, University of Reading Events 1 ICEAGE Project Curricula Development Workshop for e-Science Education Roles 0 Publications 1 Policy for Supporting Grid Education and Training Areas of interest 1 e-Science curricula development Claim 138 Awarded: £584 Claimed: £399.99 To see the full details of Prof Baker’s GridNet2 activities, see section 3.4, Prof Mark Baker – Reading. 2.5 Dr Dave Berry – NeSC Dr Berry was funded by Prof Malcolm Atkinson’s GridNet2 award (100). Table 8: GridNet2 Summary for Dave Berry No. Description People 1 Dr Dave Berry, NeSC, University of Edinburgh Events 5 GGF16, GGF18, OGSA F2F – Jan 06, OGSA Telcons – Aug 06, OGF19 Page 23 2 GridNet2 Member Summaries Roles No. Description 9 Co-chair – Grid Computing Now! The UK Experience in the adoption of Grid Computing Technologies (GGF18) Co-chair – OGSA-D-WG (GGF16, OGSA F2F – Jan 06, GGF18) Co-chair – OGSA Data Architecture Overview (GGF16) Co-chair – OGSA Data Architecture Scenarios (GGF18) Co-chair – OGSA Data Architecture Services (GGF18) Co-chair – OGSA Data Architecture Working Session (GGF16) Co-chair – OGSA Data Document Review (OGF19) Co-chair – EMS/Data Scenarios (OGF19) Publications 2 The Open Grid Services Architecture, version 1.5 OGSA Data Architecture Areas of interest 20 BYTEIO-WG (OGF19) DAIS-WG (GGF16) Data Area Meeting (GGF16) Data Movement Interface Standardisation (DMIS) BoF (GGF16) Development Tools for GT4 Service Programming (GGF18) Enterprise and Standards Summit Sessions (GGF18) Enterprise Grid Requirements (GGF18) GFS-WG (GGF16) Grid Architecture Experts Workshop (GGF16) Grid Interoperation Now (GGF16) GSM-WG (OGF19) OGSA-WG (GGF18, OGF19) OGSA and EGA Reference Model (GGF18) OGSA Workflow (OGF19) Pharmaceutical Grid Requirements (GGF18) QoS BoF (OGF19) SNIA Joint Session (GGF16) Standards All Hands: Integrating the Work of OGF (OGF19) Storage Resource Management Standard Proposal (OGF19) Topics in Identity Management (GGF18) Claim 100 See section 2.3, Prof Malcolm Atkinson – NeSC To see the full details of Dr Berry’s GridNet2 activities, see section 3.5, Dr Dave Berry – NeSC. 2.6 Prof David Chadwick – Kent Table 9: Gridnet2 Summary for David Chadwick No. Page 24 Description 2 GridNet2 Member Summaries No. Description People 1 Prof David Chadwick, University of Kent Events 9 GGF14, GGF15, GGF16, GGF17, GGF18, OGF19, OGF20, OGF21, OGF22 Roles 14 Co-chair – Authz Interoperation Demos (OGF22) Co-chair – OGSA-Authz WG (GGF14, GGF15, GGF16, GGF17, GGF18, OGF19, OGF20, OGF21, OGF22) Presenter – Dynamic Delegation of Authority Between Sites (GGF15) Presenter – GridNet2 Workshop (OGF21) Presenter – OGSA Authz WG (GGF14, OGF21) Publications 14 OGF Documents Functional Components of Grid Service Provider Authorisation Service Middleware OGSI Authorization Requirements (ID) Use of XACML Request Context to Obtain an Authorisation Decision Use of SAML to Retrieve Authorisation Credentials Use of SAML for OGSI Authorization Use of WS-Trust and SAML to Access a CVS Conference Papers Adding Support to XACML for Dynamic Delegation of Authority in Multiple Domains (10th IFIP TC6, TC11) Advanced Security for Virtual Organizations: The Pros and Cons of Centralized vs Decentralized Security Models (CCGrid 2008) Advanced Security Infrastructures for Grid Education (WMSCI 2006) Authorisation using Attributes from Multiple Authorities (WETICE 2006, Best Paper Award) Building a Modular Authorization Infrastructure (AHM 2006) Obligation for Role Based Access Control (SSNDS 07) Providing Secure Coordinated Access to Grid Services (MGC 2006) Supporting Decentralized, Security focused Dynamic Virtual Organizations Across the Grid (2nd IEEE Conference on e-Science and Grid Computing 2006) Page 25 2 GridNet2 Member Summaries Areas of interest No. Description 14 Astro RG Workshop (GGF17, OGF20) CAOPS-WG (OGF 19, OGF20, OGF21) Federated Identity (OGF19) Globus Security for the Real World (GGF18) Leveraging Site Infrastructures for Multi-Site Grids (GGF15) LoA BoF and RG (OGF19, OGF20, OGF21) OGSA Authn Charter BoF (OGF19) OGSA-Authz-WG (GGF14, GGF15, GGF16, GGF17, GGF18, OGF19, OGF20) OGSA Express Authentication Security Profile (OGF20, OGF21) OGSA Security (OGF19) SAGA Security (OGF19) Security Area Group Meeting (GGF16, OGF22) Security Talks (GGF18, OGF21) Shibboleth (GGF16, GGF18) Claim 88 Awarded: £12,600 Claimed: £12,805.11 To see the full details of Prof Chadwick’s GridNet2 activities, see section 3.6, Prof David Chadwick – Kent. 2.7 Mr Xiaoyu Chen – Brunel Table 10: GridNet2 Summary for Xiaoyu Chen No. Description People 1 Mr Xiaoyu Chen, Brunel University Events 1 OGF21 Roles 1 Co-chair UR-WG Publications 0 Areas of interest 2 Claim 128 UR-WG RUS-WG Awarded: £1,000 Claimed: £973.61 To see the full details of Mr Chen’s GridNet2 activities, see section 3.7, Mr Xiaoyu Chen – Brunel. 2.8 Mr Andrew Cooper – Oxford Mr Andrew Cooper was supported by Dr Andrew Martin’s GridNet2 award, (89). Page 26 2 GridNet2 Member Summaries Table 11: GridNet2 Summary for Andrew Cooper No. Description People 1 Mr Andrew Cooper, University of Oxford Events 1 GGF16 Roles 2 Presenter – Use Cases for Trusted Computing (GGF16) Presenter – Virtualization and the Grid (GGF16) Publications 1 Use Cases for Trusted Computing Areas of interest 2 Grid Applications of Virtualization Technologies BoF Claim 89 Trusted Computing RG For more information, see section 2.26, Dr Andrew Martin – Oxford. To see the full details of Mr Cooper’s GridNet2 activities, see section 3.26, Dr Andrew Martin – Oxford. 2.9 Dr Thierry Delaitre – Westminster Table 12: GridNet2 Summary for Thierry Delaitre No. Description People 1 Dr Thierry Delaitre, University of Westminster Events 1 GGF18 Roles 1 Presenter – Interoperability Issues for Multi-Grids (GGF18) Publications 0 Areas of interest 3 Distributed Simulation ETSI Grid BoF GIN-CG 115 Claim Awarded: £1,500 Claimed: £1,500 To see the full details of Dr Delaitre’s GridNet2 activities, see section 3.9, Dr Thierry Delaitre – Westminster. 2.10 Dr Abdeslem Djaoui – STFC Dr Abdeslem Djaoui was supported by Dr Steve Fisher’s GridNet2 awards (103 and 120). Table 13: GridNet2 Summary for Abdeslem Djaoui People No. Description 1 Dr Abdeslem Djaoui, STFC (Rutherford Appleton Laboratory) Page 27 2 GridNet2 Member Summaries No. Description Events 2 GGF15, OGF20, OGF21 Roles 2 Presenter – INFOD Base Specification (GGF15) Presenter – OGSA F2F (OGF20) Publications 1 OGSA Glossary of Terms v1.6 Areas of interest 9 Data/Compute Affinity – Focus on Data Caching (OGF21) INFOD-WG (GGF15) OGSA Glossary (OGF20) OGSA Information Services (OGF20) OGSA Naming (OGF20) OGSA WG (GGF15, OGF20) OGSA Workflow (OGF 20) Understanding Grid in the Distributed Computing Landscape workshop (OGF21) Web 2.0 Grids and Cyberinfrastructure (OGF21) Claim 103 120 See section 2.12, Dr Steve Fisher – STFC. To see the full details of Dr Djaoui’s GridNet2 activities, which are incorporated into Dr Fisher’s reports, see section 3.12, Dr Steve Fisher – STFC. 2.11 Mr Donal Fellows – Manchester Table 14: GridNet2 Summary for Donal Fellows No. Description People 1 Mr Donal Fellows, University of Manchester Events 4 OGF21, OGF22, OGSA F2Fs (Aug 07 and Jan 08) Roles 5 Chair – OGSA-RSS-WG EPS Discussion (OGF20) Co-chair – Resource Selection Discussion (OGF22) Co-chair – UR-WG (OGF20) Co-chair – Usage Record Format Working Session (OGF22) Editor for Reference Model Publications 6 Job Submission Description Language (JSDL) Specification v1.0 JSDL HPC Profile Application Extension, Version 1.0 JSDL SPMD Application Extension, Version 1.0 OGSA Glossary of Terms OGSA Resource Selection Services: Specification Web Services Agreement Specification (WS-Agreement) Page 28 2 GridNet2 Member Summaries Areas of interest No. Description 13 Activity Instance Document Schema BoF (OGF22) GLUE-WG (OGF20, OGF22) HPCP-WG (OGF20, OGF22) Information Modelling for Computing Resources workshop (OGF20) JSDL-WG (OGF20, OGF22) National and International Grid Education and Training Policy (OGF20) OGSA-RUS-WG (OGF 20, OGF22) OGSA-WG General Session (OGF22) OGSA Info/Data Modelling Architecture (OGF22) RM-WG (OGF20, OGF22) RSS Interface Specification workshop (OGF20) Towards a Usage Record 2.0 Workshop (OGF20) UR-WG (OGF20) Claim 125 Award: £5000 Claimed: £4,703.35 To see the full details of Mr Fellows’ GridNet2 activities, see section 3.11, Mr Donal Fellows – Manchester. 2.12 Dr Steve Fisher – STFC Table 15: GridNet2 Summary for Steve Fisher People No. Description 3 • Dr Steve Fisher, STFC (Rutherford Appleton Laboratory) • Abdeslem Djaoui, STFC (Rutherford Appleton Laboratory) For more information, see section 2.10, Dr Abdeslem Djaoui – STFC • Mark Leese, STFC (Daresbury Laboratory) For more information, see section 2.24, Mr Mark Leese – STFC. Events 11 INFOD-WG F2F x5, GGF15, GGF16, GGF18, OGF19, OGF20, OGF21 Roles 7 Co-chair INFOD-WG (GGF18) Presenter – Data Area Meeting (OGF19, OGF20) Presenter – INFOD-WG status (GGF16) Presenter – SAGA and Service Discovery (OGF19, OGF20, OGF21) Publications 2 Information Dissemination in the Grid Environment - Base Specifications SAGA Extension: Service Discovery API Areas of interest 2 INFOD WG SAGA Page 29 2 GridNet2 Member Summaries Claim No. Description 103 120 Award: £29,500 Claimed: £18,100.36 To see the full details of Dr Fisher’s GridNet2 activities, see section 3.12, Dr Steve Fisher – STFC. 2.13 Mr Andrew Harrison - Cardiff Table 16: GridNet2 Summary for Andrew Harrison No. Description People 1 Mr Andrew Harrison, Cardiff University Events 3 OGF20, OGF21, OGF22 Roles 5 Presenter – Embedding Workflows into Different Environments (OGF21) Presenter – Workflows Hosted in Portals (WHIP) (OGF20) Presenter – Workflow Management RG (OGF22) Secretary – WFM-RG (OGF20, OGF21) Publications 0 Areas of interest 8 e-Arts and e-Humanities (OGF20) GridNet2 e-Science Workshop (OGF21) OGSA Workflow (OGF20, OGF21) OMII Software (OGF20) Sharing Workflows (OGF20) Web 2.0 Grids and Cyberinfrastructure (OGF21) Web 2.0 Technology and Semantic Research Grid (OGF21) Workflow Management RG (OGF20, OGF21, OGF22) 122 135 Claim Award: £3,500 Claimed: £3,451.75 To see the full details of Mr Harrison’s GridNet2 activities, see section 3.13, Mr Andrew Harrison – Cardiff. 2.14 Mr Chris Higgins – Edinburgh Table 17: GridNet2 Summary for Chris Higgins No. Description People 1 Mr Chris Higgins, EDINA, University of Edinburgh Events 5 OGC – June 06, GGF18, OGF20, OGF21, OGF22 Page 30 2 GridNet2 Member Summaries Roles No. Description 8 Co-organiser - OGF-OGC Collaboration workshop (OGF20, OGF21, OGF22) Liaison between Grid/e-Science and Geographic Information (GI) communities Presenter - OGF-OGC Collaboration workshop (OGF20, OGF21, OGF22) Presenter – GridNet2 workshop Publications 0 Areas of interest 10 Astronomical Virtual Observatory (OGF20) Authz Interoperation Demos (OGF22) Earth Observation and Ground Systems (GGF18) GridNet2 workshop (OGF21) National and International Grid Education and Training Policy (OGF21) OGF-OGC Collaboration (OGF20, OGF 21, OGF22) OMII-Europe (OGF22) Security (GGF18) Security Area Meeting (OGF21) Web 2.0 Grids and Cyberinfrastructure (OGF20) Claim 101 Award: £12,000 Claimed: £10,604.01 To see the full details of Mr Higgins’ GridNet2 activities, see section 3.14, Mr Chris Higgins – Edinburgh. 2.15 Mr Neil Chue Hong – OMII-UK Mr Neil Chue Hong inherited Dr Steven Newhouse’s GridNet2 award (99) when he replaced Dr Newhouse as Director of OMII-UK. Table 18: GridNet2 Summary for Neil Chue Hong No. Description People 1 Mr Neil Chue Hong, OMII-UK Events 2 OGF21, OGF22 Roles 3 Chair NomCom (OGF21) Chair OMII-UK Workshop: Reducing the Gap between Researchers and Resources (OGF21) Co-chair OGSA-Byte I/O (OGF21) Publications 1 OGSA Byte I/O Interoperability Testing Specification Page 31 2 GridNet2 Member Summaries Areas of interest No. Description 19 Authz Interoperation (OGF22) caGrid 1.0 (OGF21) DAIS-WG (OGF21, OGF22) Data Area Meeting (OGF21) Data Management (OGF22) Financial Services (OGF22) GIN (OGF21, OGF22) Grids in the IT Data Centre (OGF21) HPCP (OGF22) OGC/OGF Collaboration (OGF21) OGSA-Byte I/O (OGF21) OGSA-DAI (OGF21) OGSA Data Architecture (OGF21, OGF22) OGSA-DMI (OGF22) OMII-Europe and OMII-UK (OGF21, OGF22) SAGA (OGF21, OGF22) Standard API for Data Grids (OGF21, OGF22) The Encyclopedia of Life (OGF22) Web 2.0 (OGF21) Claim 99 See section 2.27, Dr Steven Newhouse – OMII-UK. To see the full details of Mr Chue Hong’s GridNet2 activities, see section 3.15, Mr Neil Chue Hong – OMII-UK. 2.16 Dr Richard Hughes-Jones – Manchester Table 19: GridNet2 Summary for Richard Hughes-Jones No. Description People 1 Dr Richard Hughes-Jones, University of Manchester Events 4 GGF15, OGF19, OGF20, OGF21 Roles 7 Area Director – Infrastructure Area (OGF20, OGF21) Chair – Infrastructure Area (OGF20) Co-Chair NM-WG (GGF15, OGF19, OGF20) Presenter – Networking and Infrastructure (OGF21) Publications Page 32 0 2 GridNet2 Member Summaries Areas of interest No. Description 13 Astronomy Applications meeting (OGF19) Firewall Issues RG meeting (OGF19, OGF20, OGF21) Grids and System Virtualisation Group (OGF19, OGF20, OGF21) Grid High Performance Networking (OGF19, OGF20, OGF21) GridNet2 e-Science Workshop Infrastructure Area Meeting (OGF19, OGF21) Network Markup Language Working Group (OGF20, OGF21) Network Measurement WG (OGF19, OGF20, OGF21) Network Measurements workshop (GGF15) Particle and Nuclear Physics Applications (OGF19) Standards All Hands Meeting (OGF21) Storage Networking Community Group (OGF19) Web Services for NM Applications (GGF15) Claim 102 Award: £6,000 Claimed: £5,844.05 To see the full details ofDr Hughes-Jones’ GridNet2 activities, see section 3.16, Dr Richard Hughes-Jones – Manchester. 2.17 Dr Jens Jensen – STFC Table 20: GridNet2 Summary for Jens Jensen No. Description People 1 Dr Jens Jensen, STFC (Rutherford Appleton Laboratory) Events 1 OGF22 Roles 8 Chair – GSM-WG (OGF22) Presenter – SRM 2.2 Protocol Review (OGF22) Presenter – Data Management Issues for GSM (OGF22) Presenter – Private Key Protection (OGF22) Presenter – Certificate Renewal (OGF22) Presenter – CP/CPS Model Template (OGF22) Presenter – Robot Certificates (OGF22) Presenter – Higher Level CAs (OGF22) Publications 1 Grid Certificate Profile Page 33 2 GridNet2 Member Summaries Areas of interest No. Description 11 ASTRO-RG CAOPS-IGTF CAOPS-WG GFS-WG GIN GSM-WG PE-RG RUS-WG Storage Resource Brokers TeraGrid Access Management Workflow Engines Claim 136 Award: £1,614 Claimed: £1,354.20 To see the full details of Dr Jensen’s GridNet2 activities, see section 3.17, Dr Jens Jensen – STFC. 2.18 Dr Shantenu Jha – UCL Table 21: GridNet2 Summary for Shantenu Jha No. Description People 1 Dr Shantenu Jha, University College London Events 7 GGF15, GGF16, GGF17, GGF18, SAGA Design and Implementation Meeting, LGC Workshop, ISSGC07 Roles 6 Chair for SAGA-RG and SAGA-WG (GGF15, GGF16, GGF17, GGF18) Presenter – SAGA status (GGF15) Presenter – Introduction to SAGA (ISSGC07) Publications 3 A Collection of Use Cases for a Simple API for Grid Applications A Requirements Analysis for a Simple API for Grid Applications A Simple API for Grid Applications (SAGA) Areas of interest 8 SAGA and Grid CPR (GGF16) SAGA API (GGF16) SAGA Design (GGF17) SAGA Implementations (GGF16) SAGA Interoperability (GGF16) SAGA Outreach (ISSGC 07) SAGA Research Group (GGF15, GGF16) Lightweight Grid Computing Page 34 2 GridNet2 Member Summaries Claim No. Description 86 Award: £10,000 Claimed: £9,777.25 To see the full details of Dr Jha’s GridNet2 activities, see section 3.18, Dr Shantenu Jha – UCL. 2.19 Mr Jipu Jiang – NeSC Mr Jiang was supported by Prof Richard Sinnott’s GridNet2 award (97). Table 22: GridNet2 Summary for Jipu Jiang No. Description People 1 Mr Jipu Jiang, NeSC, Glasgow Events 1 GGF18 Roles 0 Publications 0 Areas of interest 6 Education and Training Community Group Workshop Identity Management OGSA Authz WG Security Talks Shibboleth for Grids Storage Grids in Healthcare Claim 97 See section 2.34, Prof Richard Sinnott - NeSCProf Richard Sinnott - . To see the full details of Mr Jiang’s GridNet2 activities, see section 3.19, Mr Jipu Jiang – NeSC. 2.20 Dr Mike Jones – Manchester Table 23: GridNet2 Summary for Mike Jones No. Description People 1 Dr Mike Jones, University of Manchester Events 5 GGF16, GGF17, GGF18, OGF20, OGF21 Roles 5 NGS Authz and Authn Representative for GIN Presenter – ES-LoA Project (OGF21) Presenter – Integrating Shibboleth into GridSite (GGF16) Presenter – SHEBANGS Status (GGF16, GGF18) Publications 1 Grid Certificate Profile Page 35 2 GridNet2 Member Summaries Areas of interest No. Description 24 Aggregating Mobile Devices with Grids (GGF18) CAOPS (GGF16, GGF17, GGF18, OGF20, OGF21) Data Access and Integration (GGF17) Dynamic Service Level Agreements (OGF20) Firewall Issues Research Group (GGF17, GGF18, OGF21) GIN (GGF16, GGF17, OGF20) gLite (OGF20) Grid and Shibboleth (GGF16, GGF18) Grid Authorization – Interoperability Here and Now (GGF16) Grids and Virtualisation (GGF16, OGF21) GSM-WG (OGF21) GT4 Status & Experiences, Applications and Deployments (GGF17) IGTF (GGF16, GGF18, OGF21) Information Modelling for Computing Resources (OGF20) Interoperability (GGF16) JSDL WG (OGF20, OGF21) LoA RG (OGF20, OGF21) OGF101 (GGF18) OGSA Authz (GGF16, GGF17, GGF18, OGF20, OGF21) SAGA (OGF21) Security Area Session (GGF17, OGF21)) Standard API for Data Grids (OGF21) TeraGrid Security (GGF17) Topics in Identity Management (GGF18) Claim 109 126 Award: £8,050 Claimed: £7,624.92 To see the full details of Dr Jones’ GridNet2 activities, see section 3.20, Dr Mike Jones – Manchester. 2.21 Mr Ian Kelley – Cardiff Table 24: GridNet2 Summary for Ian Kelley No. Description People 1 Mr Ian Kelley, Cardiff University Events 4 GGF18, OGF20, OGF21, OGF22 Roles 0 Publications 0 Page 36 2 GridNet2 Member Summaries Areas of interest No. Description 13 Collaborative Grids (OGF20) e-Arts and e-Humanities (OGF20) Grid: A Means to What End? (OGF20) GridNet2 Meeting (OGF21) GRIDREL-RG (OGF21) GridSphere (GGF18, OGF21) myExperiment (OGF21) OGSA Workflow (OGF20) Scaling up to the Enterprise Level (OGF20) Service Oriented Knowledge Utilities (OGF20) Sharing Workflows (OGF20, OGF22) Web 2.0 (OGF21) Web Portals (OGF21) Claim 111 132 Award: £6,000 Claimed: £5,978.07 To see the full details of Mr Ian Kelley’s GridNet2 activities, see section 3.21, Mr Ian Kelley – Cardiff. 2.22 Dr Amrey Krause – Edinburgh Table 25: GridNet2 Summary for Amrey Krause No. Description People 1 Dr Amrey Krause, University of Edinburgh Events 5 GGF16, GGF18, OGF20, OGF21, OGF22 Roles 3 Chair – Data Integration Solutions with OGSA-DAI (OGF21) Presenter – OGSA DAI Technology Update (GGF16) Presenter – OGSA-DAI Overview (OGF20) Publications 3 Web Services Data Access and Integration - The Core (WS-DAI) Specification Web Services Data Access and Integration - The Relational Realisation (WS-DAIR) Specification Web Services Data Access and Integration - The XML Realization (WSDAIX) Specification Page 37 2 GridNet2 Member Summaries Areas of interest No. Description 22 ByteIO WG (GGF16, GGF18, OGF21, OGF22) Converging Web Services Standards BoF (GGF16) DAIS-WG (GGF16, GGF18, OGF20, OGF21, OGF22) Data Area Meeting (GGF16, GGF18) Data Integration Solutions with OGSA-DAI (OGF21) Data Management (GGF18, OGF22) Data Management Convener (OGF20) DFDL-WG (OGF22) Education and Training Community Group Workshop (GGF18) Encyclopedia of Life (OGF22) Globus and Community (OGF20, OGF21) Globus committers meeting (OGF20) GridNet2 Workshop (OGF21) OGSA-DAI and Information Retrieval (OGF20) OGSA-DAI Performance Testing (OGF20) OGSA-DAI Technology Update (GGF16) OGSA-DAI User Forum/Group (GGF18, OGF20, OGF21) OGSA-DMI Working Group (GGF18, OGF20, OGF22) OGSA-D-WG (OGF22) OMII UK – Reducing the Gap Between Researchers and Resources (OGF21) RM-WG (OGFF20) SAGA (OGF20, OGF22) Claim 105 Award: £8,000 Claimed: £6,497.63 To see the full details of Dr Krause’s GridNet2 activities, see section 3.22, Dr Amrey Krause – Edinburgh. 2.23 Mr William Lee – Imperial Mr William Lee was supported by Dr Steven McGough’s GridNet2 awards (90, 121, and 134). Table 26: GridNet2 Summary for William Lee No. Description People 1 Mr William Lee, Imperial College London Events 3 GGF15, GGF16, GGF17, Roles 3 Contributor – OGSA-BES Presenter – GridSAM at JSDL-WG (GGF17) Presenter – Introduction to GridSAM (GGF16) Page 38 2 GridNet2 Member Summaries Publications No. Description 3 Grid Economy Use Cases JSDL SPMD Application Extension, v1.0 OGSA® Basic Execution Service Version 1.0 Areas of interest 7 Interoperability Fests (GGF16) JSDL Post v1.0 Discussions (GGF16) JSDL Workshop (GGF16, GGF17) OGSA EMS Architecture (GGF17) OGSA-BES General Discussion (GGF16) OMII-UK: Providing the e-Infrastructure and Tools to Enable e-Science (GGF16) SAGA Implementations (GGF16) Claim 90 For more information, see section 2.25, Dr Stephen McGough – Imperial. To see the full details of Mr William Lee’s GridNet2 activities, which are incorporated into Dr McGough’s reports, see section 3.25, Dr Stephen McGough – Imperial. 2.24 Mr Mark Leese – STFC Mr Mark Leese was supported by Dr Steven Fisher’s GridNet2 awards, claims 103 and 120. Table 27: GridNet2 Summary for Mark Leese No. Description People 1 Mr Mark Leese, STFC (Daresbury Laboratory) Events 3 GGF15, GGF17, OGF20 Roles 5 Co-Chair NM-WG (GGF15, GGF17, OGF20) Presenter – Network Measurements Basics (GGF17) Presenter – Performance Monitoring/GridMon (GGF15) Publications 0 Areas of interest 6 Firewall Issues Research Group (GGF17) Grid High Performance Networking Research Group (GGF17) GridWorld Japan (GGF17) Infrastructure Area Meeting (OGF20) Network Measurements Working Group (GGF15, GGF17) Web Services Performance (GGF15) Claim 103 120 For more information, see section 2.12, Dr Steve Fisher – STFC. Page 39 2 GridNet2 Member Summaries To see the full details of Mr Leese’s GridNet2 activities, which are incorporated into Dr Fisher’s reports, see section 3.12, Dr Steve Fisher – STFC. 2.25 Dr Stephen McGough – Imperial Table 28: GridNet2 Summary for Stephen McGough People No. Description 3 Dr Stephen McGough, Imperial College London Mr William Lee, Imperial College London For more information, see section 2.23, Mr William Lee – Imperial. Mr Vesselin Novov, Imperial College London For more information, see section 2.28, Mr Vesselin Novov – Imperial. Events 18 GGF14, GGF15/OGSA F2F, GGF16, GGF17, GGF18, OGF19, OGF20, OGF21, OGF22, OGSA F2Fs 12, 13, 14, 15, 16, 17, 18, SC06 and SC07 Roles 19 Chair – OGSA Workflow (OGF20, OGF21) Chair – QoS BoF (OGF19) Co-chair – JSDL WG (GGF14, GGF15, GGF16, GGF17, OGF19, OGF20, OGF21, OGF22) Co-chair – Activity Instance Document Schema BoF (OGF22) Co-chair – GridNet2 e-Science Workshop (OGF21) Co-chair – JSDL Activity Instance Schema (OGF22) Presenter – GGF Job Description (GGF17) Presenter – GridSAM (GGF14, GGF16, GGF17, OGF21) Publications 2 Job Submission Description Language (JSDL) Specification v1.0 JSDL SPMD Application Extension, v1.0 Areas of interest 28 CDDLM WG (GGF15) Cloud Systems BoF (OGF22) Compute Area Meeting (OGF19) DMTF Technologies (OGF19) Encyclopedia of Life (OGF22) GIN (GGF18, OGF19) GRAAP WG (GGF14, GGF15, OGF22) GridNet2 e-Science (OGF21) GSA-RG (GGF16, OGF20, OGF22) HPC Profile BoF (GGF16, GGF18) HPCP Specification Adoption (OGF22) Joint IVOA-GGF Workshop (GGF17) JSDL WG (GGF14, GGF15, GGF16, GGF17, OGF19, OGF20, OGF21, OGF22) OGSA and GLUE Information and Data Modelling (OGF19, OGF20, OGF22) OGSA Authz-WG (OGF20) Page 40 2 GridNet2 Member Summaries No. Description OGSA Security (OGF19) OGSA Workflow (OGF20, OGF21) OGSA-BES WG (GGF14, GGF15, GGF16, GGF17, OGF19) OGSA-D-WG (OGF19) OGSA-EMS WG (GGF14, GGF15, GGF17, OGF19, OGF20, OGF22) OGSA-WG (OGF19, OGF20, OGF21, OGF22) OMII-UK (GGF16,OGF19 OGF21) QoS (OGF19) RSS WG (GGF15, GGF16, GGF17, OGF21) SAGA WG (GGF14, GGF16, GGF17) Shibboleth for Grids/GridShib (GGF18, OGF20) Standards All Hands Meeting (OGF19, OGF21) WFM-RG (OGF20, OGF22) Claim 90 121 134 Award: £55,800 Claimed: £52,911.29 To see the full details of Dr McGough’s GridNet2 activities, see section 3.25, Dr Stephen McGough – Imperial. 2.26 Dr Andrew Martin – Oxford Table 29: GridNet2 Summary for Andrew Martin People No. Description 2 • Dr Andrew Martin, University of Oxford • Mr Andy Cooper, University of Oxford For more information, see section 2.8, Mr Andrew Cooper – Oxford. Events 3 GGF15, GGF16, TCG Members Meeting Roles 5 Co-chair – TC-RG (GGF16) Organiser – Grid Applications of Virtualization Technologies BoF (GGF16) Presenter – ShibGrid (GGF16) Presenter – Use Cases for Trusted Computing (GGF15, GGF16) Publications 1 Use Cases for Trusted Computing in Grids Page 41 2 GridNet2 Member Summaries Areas of interest No. Description 12 Firewall Issues RG (GGF15, GGF16) Grid Authorization – Interoperability Here and Now (GGF16) Leveraging Site Infrastructure for Multi-Site Grids (GGF15) OGSA Authz WG (GGF15) OGSA-D Security Discussion (GGF15) Shibboleth and Grid (GGF16) TC-RG: Reports and Discussion (GGF16) Trusted Computing – Enterprise Storage and Key Management (TCG Nov 06) Trusted Computing – Server Issues (TCG Nov 06) Trusted Computing – Virtualization (TCG Nov 06) Trusted Computing Authentication Technologies (TCG Nov 06) Use Cases for Trusted Computing (GGF15) Claim 89 Award: £10,000 Claimed: £5,643.91 To see the full details of Dr Martin’s GridNet2 activities, see section 3.26, Dr Andrew Martin – Oxford. 2.27 Dr Steven Newhouse – OMII-UK Table 30: GridNet2 Summary for Steven Newhouse People No. Description 4 • Dr Steven Newhouse, OMII-UK • Dr Mario Antonioletti, EPCC, University of Edinburgh For more information, see section 2.2, Dr Mario Antonioletti – EPCC. • Mr Neil Chue Hong, OMII-UK For more information, see section 2.15, Mr Neil Chue Hong – OMII-UK. • Mr Elias Theocharopoulos, NeSC, University of Edinburgh For more information, see section 2.38, Mr Elias Theocharopoulos – NeSC. Events 4 GFSG – Jan 06, OGSA F2F – Jan 06, GGF16, GGF17 Roles 9 Area Director – Application Standards (GFSG, GGF16, GGF17) Co-chair OGSA EMS Architecture (GGF17) Co-chair OGSA-BES-WG (GGF16) Co-organiser – Bridging the Divide: Community Application Requirements Driving Standards Development workshop (GGF17) Organiser – OMII-UK workshop (GGF16) Presenter – Developing with OMII (GGF16) Presenter - Joint IVOA-GGF: Capturing User Needs – A Middleware Perspective (GGF17) Page 42 2 GridNet2 Member Summaries Publications No. Description 6 A Simple API for Grid Applications (SAGA) A Requirements Analysis for a Simple API for Grid Applications Grid Economy Use Cases OGSA Basic Execution Service OGSA EMS Architecture Scenarios Web Services Agreement Specification (WS-Agreement) 17 Areas of interest Applications Area Meeting (GGF16, GGF17) Bridging the Divide: Community Application Requirements Driving Standards Development workshop(GGF17) Converging Web Services Standards BoF (GGF17) GFSG Steering Group (GGF16, GGF17) Grid and Shib Investigators Meeting (GGF16) Grid Architecture Experts workshop (GGF16) Grid Interoperability Now (GGF16, GGF17) HPC BoF (GGF17) Interoperability Fests (GGF16) Joint IVOA-GGF: Capturing User Needs – A Middleware Perspective (GGF17) OGSA BES (OGSA-WG F2F, GGF16) OGSA EMS (OGSA WG F2F, GGF17) OGSA Information Model (OGSA WG F2F, GGF16) OGSA Round Table (GGF16) OMII-UK: Providing the e-Infrastructure and Tools to Enable e-Science workshop (GGF16) SAGA RG (GGF16, GGF17) Semantic Grid 101 (GGF16) 99 Claim Award: £17,623 Claimed: £13,295.77 To see the full details of Dr Newhouse’s GridNet2 activities, see section 3.27, Dr Steven Newhouse – OMII-UK. 2.28 Mr Vesselin Novov – Imperial Mr Vesselin Novov was supported by Dr Stephen McGough’s GridNet2 awards (90, 121 and 134). Table 31: GridNet2 Summary for Vesselin Novov No. Description People 1 Mr Vesselin Novov, Imperial College London Events 7 GGF18, OGF20, OGF21, OGF22, OGSA F2F18, OGSA F2F19, SC07 Page 43 2 GridNet2 Member Summaries No. Roles 0 Publications 0 Areas of interest 15 Description Authz Interoperation Demos (OGF22) Cloud Systems BoF (OGF22) GIN Update (GGF18) GIR-WG (OGF22) HPCP-WG (OGF20, OGF21, OGF22) JSDL-WG (GGF18, OGF20, OGF21) OGSA-Authz-WG (GGF18, OGF22) OGSA-BES Specification (GGF18, OGF20) OGSA-WG (OGF22) OMII-UK: Reducing the Gap between Researchers and Resources (OGF21) OGSA EMS and Interop (OGF22) SAGA Software Solution (OGF21) SAGA-WG (OGF22) Shibboleth for Grids (GGF18) Vulnerability Assessment and Secure Coding Practices for Middleware (OGF21) Claim 90 121 134 For more information, see section 2.25, Dr Stephen McGough – Imperial. To see the full details of Mr Novov’s GridNet2 activities, which are incorporated into Dr McGough’s reports, see section 3.25, Dr Stephen McGough – Imperial. 2.29 Dr Savas Parastatidis - Newcastle Table 32: GridNet2 Summary for Savas Parastatidis No. Description People 1 Dr Savas Parastatidis, University of Newcastle Events 1 GGF14 Roles 0 Publications 1 Page 44 Web Services Data Access and Integration - The Core (WS-DAI) Specification, Version 1.0 2 GridNet2 Member Summaries Areas of interest No. Description 4 OGSA OGSA-BES OGSA-DAIS Web services 85 Claim Awarded: £1,750 Claimed: £1,639.57 To see the full details of Dr Parastatidis’ GridNet2 activities, see section 3.29, Dr Savas Parastatidis – Newcastle. 2.30 Dr Colin Perkins – Glasgow Table 33: GridNet2 Summary for Colin Perkins No. Description People 1 Dr Colin Perkins, University of Glasgow Events 9 IETF64, IETF65, IETF66, IETF67, IETF68, IETF69, IETF70, SUMOVER Nov 05, SUMOVER Apr 07 Roles 23 Chair – Self Address Fixing Evolution BoF (IETF70) Co-chair AVT WG (IETF64, IETF65, IETF66, IETF67, IETF68, IETF69, IETF70) Co-chair MMUSIC WG (IETF64, IETF65, IETF66, IETF66, IETF67, IETF68, IETF69) Presenter – Activities in the IETF (SUMOVER 05, SUMOVER07) Presenter – Application Performance using TFRC in Real-World Networks (IETF66) Presenter – Multiplexing RTP and RTCP on a Single UDP Port for Ease of NAT Traversal (IETF67) Presenter – Scaling Multimedia Conferencing (SUMOVER 05) Presenter – Transporting Real-Time Audio/Visual Data over DCCP (IETF64) Presenter – Update to RTP over DCCP Specification (IETF66) Presenter – Using RTP with DCCP To Adapt Multimedia Applications to Network Congestion (IETF67) Publications 3 RTP and Datagram Congestion Control Protocol Multiplexing RTP Data and Control Packets on a Single Port SDP: Session Description Protocol Areas of interest 23 AVT WG Sessions (IETF64, IETF65, IETF66, IETF67, IETF68, IETF69, IETF70) Congestion Control for Real-Time Streaming (IETF68) DCCP Working Group (IETF66) Defining Signalling Protocols with SVC Extensions (IETF65, IETF66, IETF68) Page 45 2 GridNet2 Member Summaries No. Description Developing Network Transport Protocols for eVLBI Data (IETF67, IETF68) Forward Error Correction Codes (IETF65) Harmonising SIP and RTSP (IETF65) ICE Methodology for NAT Traversal (IETF66, IETF67, IETF68, IETF69) Keep-Alive Mechanisms for NAT Traversal (IETF69) Management of RTP (IETF69) Matching ZRTP and DTLS to the RTP Model (IETF67) Media Capability Negotiation Framework (IETF68) MMUSIC Sessions (IETF64, IETF65, IETF66, IETF67, IETF68, IETF69) RTP and DCCP (IETF68) RTP Management Framework (IETF65) RTP Payload Format (IETF66, IETF69, IETF70) RTPSEC BoF (IETF66) Secure Negotiation of Encryption and Authentication Keys (IETF67) Securing RTP (IETF69, IETF70) Self Address Fixing Evolution (IETF70) Transport Area Directorate (IETF70) Using Non-Compound RTCP for Congestion Control (IETF69, IETF70) Video Codec Control Messages (IETF66, IETF68) Claim 96 129 Awarded: £13,000 Claimed: £12,993.41 To see the full details of Dr Perkins’ GridNet2 activities, see section 3.30, Dr Colin Perkins – Glasgow. 2.31 Dr Stephen Pickles – Manchester Table 34: GridNet2 Summary for Stephen Pickles No. Description People 1 Dr Stephen Pickles, University of Manchester Events 11 GGF15, GFSG F2F Jan 06, GGF16, GGF17, GFSG Jul 06, GGF18, ETSI at CoreGrid Conference, GFSG Dec 06, OGF19, OGF GIN F2F, OGF20 Page 46 2 GridNet2 Member Summaries Roles No. Description 15 Chair Local Organising Committee – OGF20 and EGEE User Forum (OGF20) OGF Area Director for Compute Area (GGF15, GGF16, GG17, GGF18, OGF19, OGF20, GFSG Jan 06, GFSG Jul 06, OGF GIN F2F) OGF Area Director for Scheduling and Resource Management OGF Liaison to ETSI – ETSI at CoreGrid Conference Organiser – Computational Steering workshops (OGF20) Organiser – Workshop on Heuristics for Implementing Semantic Knowledge Yardsticks BoF (OGF20) University of Manchester Contact for OASIS – OASIS WSRF Technical Committee Publications 8 A Simple API for Grid Applications (SAGA) HPC Basic Profile Version 1.0 JSDL SPMD Application Extension OGSA Basic Execution Service Version 1.0 OGSA Resource Selection Services Usage Scenarios for a Grid Resource Allocation Agreement Protocol Use-Cases and Requirements for Grid Checkpoint and Recovery Web Services Agreement Specification (WS-Agreement) Areas of interest 22 Compute Area Meeting (GGF17) Federated Identity (OGF19) GGF/EGA merger – GFSGF2F Jan 06 GRAAP-WG (GGF16, GGF17) Grid Interoperation Now (GGF16, GGF17, GGF18, GGF19, GIN F2F 07) Grid Management (GGF18) Grids in the Greenfield Regions (GGF16) HPC Profile BoF/ HPC WG (GGF17, GGF18, OGF19) Interoperability Fests (GGF16) JSDL-WG (GGF16, GGF17, OGF19) OGSA EMA Architecture (GGF17) OGSA Grid Infrastructure (OGF19) OGSA Security (OGF19) OGSA-BES Status (OGF19) OGSA-RSS-WG (GGF16, GGF18) OGSA-RUS Specification (OGF19) Security Talks (GGF18) Semiconductor/EDA Grid Requirements (GGF18) Shibboleth (GGF16, GGF18) Standards (GGF17, GFSG F2F Jul 06, OGF19) UR-WG (GGF16, GGF17, GGF18) Visualization on the Grid (OGF19) Page 47 2 GridNet2 Member Summaries Claim No. Description 98 Award: £9,120 Claimed: £6,902.99 To see the full details of Dr Pickles’ GridNet2 activities, see section 3.31, Dr Stephen Pickles – Manchester. 2.32 Prof Omer Rana – Cardiff Table 35: GridNet2 Summary for Omer Rana No. Description People 1 Prof Omer Rana, Cardiff School of Computer Science, Cardiff University Events 7 GGF16, GGF17, GGF18, OGF19, OGF20, OGF21, OGF22 Roles 11 Chair – Semantic Grid workshops (GGF16) Co-organiser for the following: • WS-Agreement Workshop (co-located with GGF18) • Dynamic Service Level Agreements Workshop (OGF20) • GridNet2 e-Science Workshop (OGF21) For more information, see section D.1, GridNet2 e-Science Workshop (OGF21). Presenter – CATNETS project at GRAAP WG (GGF16) Presenter – Dynamic SLAs at GRAAP WG (GGF17) Presenter – G-QoSM project (OGF19) Presenter – GRAAP Why Do We Need SLAs? (OGF21) Presenter – Healthcare@Home (GGF18) Presenter – Workflow Optimization Sharing and Using Performance Information (OGF20) Presenter – WS-Agreement Use in CATNETS (WS-Agreement Workshop) Publications 0 Areas of interest 26 Aggregating Mobile Devices with Grids (GGF18) Certificate Authorities (GGF17) Cloud Computing BoF (OGF22) Data Management Workshop (OGF22) Enterprise Adoption Workshop (OGF22) Evolution of Grids Towards Service-Oriented Knowledge Utilities (OGF20) GRAAP WG (GGF16, GGF17, GGF18, OGF19, OGF20, OGF21, OGF22) Grid Technologies for Arts, Humanities and Social Science CG (OGF21) GridNet2 e-Science Workshop (OGF21) Gridsphere Portlet Container for Portals (OGF19) Joint IVOA –GGF (Astro RG) (GGF17) JSDL (GGF17, OGF19, OGF21) Page 48 2 GridNet2 Member Summaries No. Description OGSA and EGA Reference Model (GGF18) OGSA Information Model (GGF17, GGF18, OGF19) OGSA Workflow (OGF19 OGF20) OMII Core Grid Infrastructure (OGF19) Production Grids Enterprise and Research (GGF16) Provenance Challenge Workshop (GGF18) Quality of Service BoF (OGF19) Reliability and Robustness in Grid Computing Systems (OGF19) SAGA API (OGF20) Semantic Grid (GGF16) Semantic Web (OGF19) Service Level Terms for OGSA-ByteIO and OGSA-BES (OGF20) SLA and Dynamic SLA (GGF16, GGF17, OGF20) WS Agreement (GGF16, GGF17, GGF18) Claim 107 133 Awarded: £7,800 Claimed: £7,375.97 To see the full details of Prof Rana’s GridNet2 activities, see section 3.32, Prof Omer Rana – Cardiff. 2.33 Dr Matthew Shields – Cardiff Table 36: GridNet2 Summary for Matthew Shields No. Description People 1 Dr Matthew Shields, Cardiff University Events 2 OGF20, OGF21 Roles 4 Co-secretary for WFM-RG (OGF20, OGF21) Presenter – Developing Workflow Sharing and Interoperability Use Cases (OGF21) Presenter – GridNet2 Activities for WFM-RG (OGF21) Publications 0 Areas of interest 8 Astronomical Virtual Observatory (OGF20) EGEE Workflow Convener meetings (OGF20) EU Presentation (OGF20) GridNet2 e-Science Workshop (OGF21) myExperiment (OGF21) OGSA Workflow Sharing (OGF21) Web 2.0 (OGF21) Workflow Management Research Group (OGF20, OGF21) Page 49 2 GridNet2 Member Summaries Claim No. Description 124 Awarded: £2,000 Claimed: £2,000 To see the full details of Dr Shields’ GridNet2 activities, see section 3.33, Dr Matthew Shields – Cardiff. 2.34 Prof Richard Sinnott - NeSC Table 37: GridNet2 Summary for Richard Sinnott People No. Description 4 • Prof Richard Sinnott, NeSC, University of Glasgow • Oluwafemi Ajayi, NeSC, University of Glasgow See section 2.1, Mr Oluwafemi Ajayi – NeSC. • Jipu Jiang, NeSC, University of Glasgow See section 2.19, Mr Jipu Jiang – NeSC. • Anthony Stell, NeSC, University of Glasgow See section 2.36, Mr Anthony Stell – NeSC. Events 6 GGF15, GGF16, GGF18, OGF20, OGF21, OGF22 Roles 10 Chair OGSA-DAI User Group (GGF18) Co-chair – Authz Interoperation Demos (OGF22) Organiser – OGSA-DAI User Forum (OGF20, OGF21) Presenter – Authz Interoperation Demos (OGF22) Presenter – GGF/OGF and the Future/GridNet3 (OGF21) Presenter – Shib+Grid Work at NeSC, Glasgow (GGF16, GGF18) Presenter – Shibboleth Protection and Management of Workflows (OGF20) Presenter – Use of Middleware in nanoCMOS (OGF21) Publications 3 Conference Documents Advanced Security for Virtual Organizations: The Pros and Cons of Centralized vs Decentralized Security Models (CCGrid 2008) Advanced Security Infrastructures for Grid Education (WMSCI 2006) Supporting Decentralized, Security focused Dynamic Virtual Organizations across the Grid (2nd IEEE Conference on e-Science and Grid Computing) Areas of interest 36 Authz Interoperation (OGF22) caGrid (GGF16, OGF21) CA-OPS-WG (GGF16) DAIS-WG (OGF21) Data Management Workshop (OGF22) Dynamic Agreements and Negotiation - GRAAP (GGF18) Encyclopedia of Life (OGF22) Enterprise Grid Requirements (GGF18) Page 50 2 GridNet2 Member Summaries No. Description Entertainment and Digital Content (GGF15) e-Science and CyberInfrastructure (GGF15) Financial Services Workshop (OGF22) Grid Education and Training (GGF16) Grid Interoperability (OGF21) Grid Usage and Productivity (OGF22) GridNet2 workshop (OGF21) Leveraging Site Infrastructure for Multi-Site Grids (GGF15) LSG-RG (GGF16, OGF21) OGSA Data Architecture (OGF21) OGSA-Authz WG (GGF15, OGF21, OGF22) OGSA-DAI User Forum (OGF20, OGF21) OGSA-RUS Specification (OGF22) OMII Europe and UK (OGF21) Pharma, Biotech and Life Sciences Workshop (OGF22) Pharmaceutical Grid Requirements (GGF18) SAGA Overview (OGF20) Security Area Meeting (GGF16, GGF18, OGF21, OGF22) Semantic Grid (GGF16) Semiconductor/EDA Grid Requirements (GGF18) Shibboleth (GGF16, GGF18) Standards All Hands Meeting (OGF21) Unified Grid Logging and Security Auditing (OGF20) VPMAN Project (OGF20) Web 2.0 Social Networking for MSI Researchers (OGF21) What is the Software Licensing Model for Grids (GGF15, OGF20) Why Leading IT Organizations Are Adopting Grids (GGF15) Workflow Sharing (OGF20) Claim 97 117 Awarded: £16,000 Claimed: £14,106.08 To see the full details of Prof Sinnott’s GridNet2 activities, see section 3.34, Prof Richard Sinnott – NeSC. 2.35 Mr David Spence – STFC Table 38: GridNet2 Summary for David Spence No. Description People 2 Mr David Spence, STFC (Rutherford Appleton Laboratory) Events 1 GGF18 Page 51 2 GridNet2 Member Summaries No. Description Roles 1 Presenter – ShibGrid Progress Publications 0 Areas of interest 6 Firewall Issues RG IGTF Issues OGSA Authz WG Security Talks Topics in Identity Management Towards Worldwide Grid User Support Claim 114 Awarded: £4,500 Claimed: £4,478.07 To see the full details of Mr Spence’s GridNet2 activities, see section 3.35, Mr David Spence – STFC. 2.36 Mr Anthony Stell – NeSC Mr Stell was supported by Richard Sinnott’s GridNet2 award (97). Table 39: GridNet2 Summary for Anthony Stell No. Description People 1 Mr Anthony Stell, NeSC, Glasgow Events 1 GGF16 Roles 0 Publications 2 Conference Documents Advanced Security for Virtual Organizations: The Pros and Cons of Centralized vs Decentralized Security Models (CCGrid 2008) Advanced Security Infrastructures for Grid Education (WMSCI 2006) Areas of interest 7 Grid Authorization Interoperability workshop Grid Education and Training workshop GT4 Status and Experiences OGSA Authz WG OGSA-DAI Technology and Update Security and Privacy Needs for Health Grids Shibboleth Claim 97 For more information, see section 2.34, Prof Richard Sinnott - . To see the full details of Mr Stell’s GridNet2 activities, see section3.36, Mr Anthony Stell – NeSC. Page 52 2 GridNet2 Member Summaries 2.37 Dr Ian Taylor – Cardiff Table 40: GridNet2 Summary for Ian Taylor No. Description People 1 Dr Ian Taylor, Cardiff University Events 4 GGF18, OGF20, OGF21, OGF22 Roles 9 Chair for the APP-AGG-RG (GGF18) Co-chair – WFM-RG (OGF20, OGF21, OGF22) Co-organiser GridNet2 workshop (OGF21) Organiser for the APP-AGG-RG workshop (GGF18) Presenter – Current Status of WFM-RG (OGF22) Presenter – Introduction GridNet2 e-Science workshop (OGF21) Presenter – Sharing Workflows (OGF20) Publications 0 Areas of interest 11 SAGA WG (GGF 18) EU Funding Round Table (GGF18) OGSA WG (GGF18) Life Science Grid RG (GGF18) OGSA Workflow (OGF20) WFM-RG (OGF20, OGF22) e-Arts and e-Humanities – e-Science Technologies and Methodologies in Arts and Humanities Research (OGF20) Evolution of Grids Towards Service Oriented Knowledge Utilities (OGF20) Sharing Workflows and Interoperability (OGF21) Web 2.0 Meets Grids (OGF21) e-Social Science: our Spaces (OGF21) 108 131 Claim Awarded: £4,500 Claimed: £4,496.30 To see the full details of Dr Taylor’s GridNet2 activities, see section 3.37, Dr Ian Taylor – Cardiff. 2.38 Mr Elias Theocharopoulos – NeSC Mr Elias Theocharopoulos was funded by Dr Steven Newhouse’s GridNet2 award (99). Table 41: GridNet2 Summary for Elias Theocharopoulos People No. Description 1 Mr Elias Theocharopoulos Page 53 2 GridNet2 Member Summaries No. Description Events 1 OGF22 Roles 0 Publications 0 Areas of interest 11 Data Management Workshop Encyclopedia of Life Infrastructure Management – eBay OGC-OGF Collaboration Workshop OGSA Authz (OGF22) OGSA Data Architecture – Future Directions Porting Applications with Globus Gridway SAGA – The Simple API for Grid Applications SAGA Java Language Binding SAGA+DAIS: Next Steps WFM-RG Claim 99 See section 2.27, Dr Steven Newhouse – OMII-UK. To see the full details of Mr Theocharopoulos’ GridNet2 activities, see section 3.38, Mr Elias Theocharopoulos - NeSC. 2.39 Dr David Wallom – Oxford Table 42: GridNet2 Summary for David Wallom No. Description People 1 Dr David Wallom, Oxford University e-Research Centre Events 3 GGF18, OGF20, OGF22 Roles 6 Co-chair – 2nd International Workshop on Campus and Community Grids, continuing interoperability (OGF20) Co-chair – Topics in Grid Management (GGF18) Co-chair – Software Licensing for Grids (OGF20) Panel Participant – The Astronomical Virtual Observatory – Building Operational Services on Pervasive Grids: Standards in Use (OGF20) Presenter – Opening Session: OGF-Europe (OGF22) Presenter – University of Oxford Campus Grid (OGF20) Publications 2 Workshop on Grid Applications: From Early Adopters to Mainstream Users (ID) 2nd International Workshop on Campus and Community Grids (ID) Page 54 2 GridNet2 Member Summaries Areas of interest No. Description 22 Application Domains (OGF20) Astronomical Virtual Observatory (OGF20) Campus Grids RG BoF and Workshop ( OGF20) Cloud Systems BoF (OGF22) Creating a Standard Software API for Data Grid Management Systems BoF (OGF22) Daonity: Trusted Computing Enhanced GSI (GGF18) Data Management (OGF20, OGF22) Encyclopedia of Life (OGF22) Enterprise and Standards Summit (GGF18) Financial Services Workshop (OGF22) GIN-CG (GGF18, OGF22) GIR-WG (OGF22) Grid Management (GGF18) HPC Profile (GGF18) HPCBP Specification Adoption (OGF22) OGF Marketing (OGF22) OMII-Europe (OGF22) PGS-RG (GGF18) Rule-Based Preservation Environments (OGF22) Software Licensing (OGF20) Storage Grids in Healthcare (GGF18) UR-WG (GGF18, OGF20) Claim 110 113 Awarded: £6,300 Claimed: £2,824.88 To see the full details of Dr Wallom’s GridNet2 activities, see section 3.39, Dr David Wallom – OeRC. 2.40 Dr Viktor Yarmolenko – Manchester Table 43: GridNet2 Summary for Viktor Yarmolenko No. Description People 1 Dr Viktor Yarmolenko, School of Computer Science, University of Manchester Events 1 SLA F2F Roles 2 Organiser – SLA F2F Presenter – Open Issues in SLAs for Workflows Publications 0 Page 55 2 GridNet2 Member Summaries Areas of interest No. Description 2 SLAs Workflows Award: £1,000 Claimed: £993.44 Claim To see the full details of Dr Yarmolenko’s GridNet2 activities, see section 3.40, Dr Viktor Yarmolenko – Manchester. 2.41 Mr Stefan Zasada – UCL Table 44: GridNet2 Summary for Stefan Zasada No. Description People 1 Mr Stefan Zasada, Centre for Computational Science, University College London Events 1 OGF20 Roles 0 Publications 0 Areas of interest 9 EGEE Applications GIN-CG JSDL-WG OGSA EMS Information Service OGSA-BES-WG OGSA-RSS-WG SAGA Vendor and Developer Adoption of OGF Standards Workflow Sharing Claim 118 Award: £3,000 Claimed: £2,459.70 To see the full details of Mr Zasada’s GridNet2 activities, see section 3.41, Mr Stefan Zasada – UCL. 2.42 Dr Ning Zhang – Manchester Table 45: GridNet2 Summary for Ning Zhang No. Description People 1 Dr Ning Zhang, University of Manchester Events 2 OGF19, OGF20 Page 56 2 GridNet2 Member Summaries Roles No. Description 4 Chair – LoA BoF (OGF19) Co-chair – LoA RG (OGF20) Organiser – LoA BoF (OGF19) Presenter – Federated Identity workshop (OGF19) Publications 0 Areas of interest 7 Federated Identity (OGF19) LoA BoF and RG (OGF19, OGF20) OGSA Authz WG (OGF19) OGSA Express Authn Security Profile (OGF20) OGSA Security (OGF19) SAGA Security (OGF19) Security Area Meeting (OGF19) Claim 123 Awarded: £2,000 Claimed: £2,000 To see the full details of Dr Zhang’s GridNet2 activities, see section 3.42, Dr Ning Zhang – Manchester. Page 57 3 GridNet2 Reports 3 GridNet2 Reports This section lists the reports submitted by those who applied for and received GridNet2 funding. Reports are listed in alphabetical order by the last name for the person who received the GridNet2 funding award. Some awards covered funding for more than one person. In this case, if each person covered by the award has submitted a report, reports are listed for each person covered by the award. If only one person has submitted reports covering all recipients, the reports are listed under the submitter’s name only. 3.1 Mr Oluwafemi Ajayi – NeSC Mr Oluwafemi Ajyai is a PhD student at NeSC, Glasgow. For a summary of Mr Ajayi’s GridNet2 activities, see section 2.1, Mr Oluwafemi Ajayi – NeSC. Mr Ahjayi was covered by Prof Richard Sinnott’s GridNet2 award. For more information, see section 2.34 Prof Richard Sinnott - NeSC. 3.1.1 GGF16 More participation is required to complete in earnest the objectives of the workgroup. Mr Ajayi also recommended existing solutions should be further investigated with respect to the existing agreed usecases. Solutions like ShibGrid, the interoperability of VOMS and Shibboleth should be further investigated. Lastly, attributes and metadata that are common to the Grid community need to be deliberated to achieve security interoperability. Grid Interoperation: The meeting focused on how interoperation and interoperability can be achieved on the Grid. Four areas considered for interoperability included: Summary • Security For Mr Ajayi, GGF 16 was largely a success, but he stated that more cooperation, collaboration and commitment are required from group members if any meaningful achievements are to be seen. It is recommended that task reports from previous GGF meetings should be required and presented in GGF meetings. Similarly, schedules, where possible, should be made for tasks that were agreed upon during meetings, as there seems to be a lack of continuity and commitment from previous meetings in some workgroups. • Jobs submission • Data management • Resource discovery Specific recommendations for group meetings Mr Ajayi attended are included in this report. The recommendations range from attributes/schema mappings to refined and clear use-cases for OGSADAI, ShibGrid and LSG-WG. OGSA-Authz-WG The meeting was chaired by David Chadwick and some significant progress was made; mostly the need for a CVS. Distinctions were made between the authenticity and validity of credentials. Similarity between Microsoft’s STS and CVS were pointed out, suggesting a need for validation standards in the Grid community. The meeting also saw the call for context parameters to augment existing agreed security parameters: subject, action, target and environment. The need for context is easily emphasised in a dynamic environment where security requirements for access roles changes from time to time. Although a consensus was not reached for the inclusion of context attributes at the meeting, Mr Ajayi supports it and believes it is an attribute that still needs to be considered and included in the final workgroup release. Page 58 It was noted that significant progress is being made in the area of interoperation while a lot needs to be done to achieve interoperability. Interoperation exists where multiple machines from multiple vendors communicate with one another, which at present is being achieved through commonly agreed specifications and procedures. Interoperability is currently being seen as the ability of software and Grid applications working together seamlessly. Interoperability is deemed to be achievable where agreements on tools and framework can be reached. All four interoperability areas were discussed, but the security area was the one that drew Mr Ajayi’s attention. Security issues regarding authentication and identity management were discussed, as follows: • Certificate recognition • Missing gridmap file entries • Cross boundary validation • Trust and security cracks despite common GSI and PMAs Similarly, authorisation issues were deliberated upon. It was recommended that authorisation services should not be too close to services as it currently is. It was also suggested that SAML should be considered in future upgrades of existing technologies, which will go a long way in improving interoperability. 3 GridNet2 Reports 3.1 Mr Oluwafemi Ajayi – NeSC ShibGrid The ShibGrid meeting focused on the interoperability of Shibboleth with existing Grid security infrastructures like VOMS, CAS and PERMIS. The meeting was chaired by Von Welch. The meeting also saw the presence of an Internet 2 development team member which guided the group on forthcoming Shibboleth releases and functionalities. Shibboleth issues, such as identity provider discovery and service delegation that makes interoperability with existing Grid applications, were discussed and listed for further investigations against the next GGF meeting. Other issues listed for investigation against next meeting included: refined VO definition, VO management interoperability, names mapping, shibportal architecture and VO-VO federation. Names mapping, shib-portal architecture and VO-VO federation recommendations were being investigated in Glasgow and the team in Glasgow continued to research and deploy models to test and resolve these issues. The existing Glasgow projects where these recommendations are being investigated include ESP-Grid, VOTES, BRIDGES and DyVOSE. CAOPS-WG The main focus in this meeting was on augmenting the proposed standard to support revocation of proxy credentials without resorting to the brute method of revoking the user certificate. Other issues on OCSP draft were discussed, such as conformance of other standards to the proposed OCSP and the time latency for certificate revocation referenced to the draft. LSG-WG There was little progress made in the Life Science Group meeting because interest in the group vision was lacking by the attendees. In Mr Ajayi’s opinion, this was probably due to exhaustion from previous meetings. There was also little continuation or feedback from previous meetings to this meeting. The reason for this was not entirely clear to Mr Ajayi as this was his first GGF meeting. However, from the deliberation that occurred, it appeared that the workgroup requirements were not understood by majority of the participants. It was recommended by a caGrid member and a member of Mr Ajayi’s VOTES team that use cases for Healthcare & Life Science security requirements should be shared on the mailing list before the next GGF meeting. Conclusions were reached and Richard Sinnott of the VOTES team agreed to make use cases from VOTES project available before the next GGF meeting. The team will be making the uses cases available over the next few weeks. 3.1.2 GGF18 Mr Ajayi attended participated in the community and standards sessions at GGF18. Enterprise Grid Requirements interoperability issues between organisations from a security perspective. The session kicked off with reviews from work being done by the working groups. The working groups include the following groups • Reference model • Computing provisioning • Data provisioning • Security • Utility accounting To facilitate interoperability, a reference model that enables enterprise partners to interoperate was discussed. The group introduce a Grid management entity (GME) which is a collection of interconnected (network) Grid components. As part of GME a reference model comprised of resource provisioning, management and deployment was highlighted. It seemed obvious that ontologies will be an important part of inter-enterprise collaboration. Similarly, some use cases for the reference model were considered and proposed. Shibboleth for Grids In the second session, the first talk was given by Richard Sinnott on the use of Shibboleth on DyVose and GLASS projects at Glasgow University. Another project Shibbolised at Glasgow was the VOTES project which Mr Ajayi is actively involved in. The talked showed how Grid resources could be Shibboleth-enabled. The second talk was by Mike Jones from Manchester University and he gave a detailed architectural view of how they have implemented Shibboleth and their experiences with Shibboleth in the SHEBANGS project. Some implementation issues as regards identity assertions were raised and also on how certificates exchanged using Shibboleth could be made into a GSI certificates. Other issues were raised such as mod_shib exported environment, attributes/assertion reuse and inadequate Shibboleth configuration information. Another presentation was by David Chadwick who gave a demonstration of policy wizard and policy editor for the creation of deployable policies by developers or non-developers alike. The relevance of this presentation was not immediately understood in the context of Shibboleth but was later clarified with further discussions in the context of exchanging assertions between PERMIS protected resources. Nathan Klingenstein from Internet2 gave an update of shibboleth developments at Internet2 and what should be expected in the next Shibboleth releases. Another talk was on X.509 and SAML 2.0, which covered how SAML/Shibboleth assertions could be translated to X.509 like DNs and attributes (that is, Shibboleth to VOMS and vice versa). Some of the noted issues at the end of the sessions included: Shibboleth-CA requirements; binding SAML to X.509; and need for attribute vocabularies (for example, names & namespaces for GridPerson). The session was a follow up of the work done since the last GGF, which focused on carefully crafting requirements that would support interoperability at enterprise levels. The first technical work done so far was released in July and a subsequent update was due imminently and would cover enterprise Grid security such as threats/risks, use cases and requirements. Mr Ajayi found the session to be of value as the focus of his work at NeSC involved Page 59 3 GridNet2 Reports 3.2 Dr Mario Antonioletti - EPCC Grid Resource Allocation and Agreement Protocol (GRAAP) WG – Session 3 Mr Ajayi joined this session because of his current research interest in federated trust negotiations on the Grid. True to the set objective, the session gave the current status of dynamic agreements and negotiation in GRAAP. The workgroup focused on resource negotiation as opposed to security credentials; however, credentials are indeed resources that need to be negotiated, hence the relevance. The workgroup was a step away from WS-Agreement but it was noticeable that little was understood on the subject area and on how to turn it into a specification. Most of the discussions were on how to modify an agreement after an agreement has been reached. Notable points from the discussion were: what modifications might be allowed; when they could be allowed; and what should trigger the modifications. In all, dynamic agreements remain an active area of interest and some use cases / requirements are still being developed. Mr Ajayi found the life science research group session particularly noteworthy because of the enormous quantity of work being done in storage Grids as regards healthcare and life sciences. Based on his last LSG-RG participation at GGF16 in Athens, the LSGRG at GGF18 was more rewarding as, for example, Richard Bakalar presented IBM’s experiences with life science and healthcare solutions, which led on to further discussion. Bakalar emphasised requirements such as high availability, data integrity, reliability, and collaboration through health care teams. The key lesson from the IBM experience was the use of fixed content over IP networks in order to achieve high availability and reliability of data. The BYCAST solution was also presented in the session. BYCAST also uses fixed content techniques over IP networks to deliver availability and dependability. The availability and reliability techniques learnt from these LSG-RG sessions will be explored in the VOTES project that Mr Ajayi is working on at NeSC. All in all, Mr Ajayi found this a useful OGF meeting and expects to continue to participate actively in the above groups. OGF20 Mr Ajayi attended the following key sessions/forum at OGF20: • Grids Means Business • Dynamic Service Level Agreement • OGSA-DAI User Forum The keynote speech delivered by Tony Heys emphasised some of the trust management work that Mr Ajayi and his team have been doing at NESC. In his speech, Heys talked about how customers want services like Amazon ID, Google ID, and MS Passport to interoperate so that data can be accessed and shared. This work area remains important as the open community continues to develop open standards such as SAML, Shibboleth and XACML. The OGSA-Authz group is currently investigating interoperability between these standards. Page 60 At the first session of Grid Means Business, a speaker from eBay presented some architectural lessons based on their Grid-integrated platforms. The talk shows how Grid applications could be developed to better meet business needs. One lesson concerned the importance of minimising dependencies during application development. Another lesson was about resilience and scalability, and virtualisation was suggested as a method of providing this. At the security session of Grid Means Business, it was argued that security management in Grids remains a challenge. Mike Boniface discussed the need for site independence in VO security management. For example, when a site is compromised, every other site should be able to decide independently how they want to mitigate the risk. At the session, organisational structure and legal frameworks were revealed as some problems facing federation (WSfederation, WS-trust, WS-policy, and WSexchange). Dynamic Service Level Agreements Workshops LSG-RG 3.1.3 Grids Mean Business Owing to its relevance to Mr Ajayi’s work at NeSC, he attended the dynamic service level agreement workshops. At the meeting, WS-Agreement for negotiating service level agreement (SLA) was discussed. Various scenarios and requirements were presented and discussed, such as those from OntoGrid, NextGrid and Alcatel-Lucent. The meeting raised questions such as credential negotiation and negotiation strategies. This aspect of negotiation ties with Mr Ajayi’s work on trust negotiation at NeSC. OGSA-DAI User Forum Lastly, as part of the Glasgow NeSC team, Mr Ajayi gave a talk at OGSA-DAI user forum on his experience with OGSA-DAI. At the forum, Mr Ajyai discussed some of the improvements that the team at NeSC would like to see it the next OGSA-DAI release., Some of the suggestions are already planned for next release. In all, OGF 20 was useful for Mr Ajayi. 3.2 Dr Mario Antonioletti - EPCC Dr Mario Antonioletti is Principal Consultant at the EPCC. Dr Mario Antonioletti was funded by Dr Steven Newhouse’s GridNet2 award. For a summary of Dr Antonioletti’s GridNet2 activities, see section 2.2, Dr Mario Antonioletti – EPCC. 3.2.1 OGF22 Dr Antonioletti attended the following sessions at OGF22: • Opening Session • SAGA, Java Language Bindings • OMII-Europe: Using open standards to deliver interoperability • OGSA-DMI Discussion of Specification 3 GridNet2 Reports 3.2 Dr Mario Antonioletti - EPCC • OGSA-Data Architecture Future Directions OGSA-D-WG • DFDL-WG Progress Report • Keynote: What OGF Can Do For Enterprises • HPCP Specification Adoption • GIN-CG: Specification Adoptions & Discussions • Globus Software User Experiences Dr Antonioletti was also involved in the OGSA-D-WG which produced an informational document last December (GFD.121) which fleshed out the data part of OGSA. Both of the WG chairs – Dave Berry and Allen Luniewski – resigned, having completed what the group charter set out to do. Nevertheless the group felt there were other tasks that needed to be considered. For more information, see: • SAGA + DAIS: Next Steps • Creating a standard software API for Data Grid Management Systems BoF • The Encyclopedia of Life: A Web page for every species • SRM 2.2 protocol review • Keynote: Cloud Computing, Grids and the upcoming Cambrian Explosion in IT • Cloud Systems BoF (1/2) • DAIS-WG Session • Town Hall Part 2 • Grid Usage and Productivity in HPC • Data Management Workshop SAGA Java API Dr Antonioletti attended the SAGA Java API, in part because some of this work was funded by OMII-UK. It also had some relevance to the DAIS-WG, for which he was co-chair, and a DAIS joint meeting that was to be held at OGF later on in the week. SAGA, if successful, would provide a strong incentive for folks to use the Grid and as such was important for all concerned. OMII-Europe Dr Antonioletti attended the OMII-Europe session as he had been previously been involved in this project – sadly the project did not appear to be getting a continuation in funding just as it was beginning to make an impact. The objective was to try and maintain some of the infrastructure in case the project did get some continuation funding. There are a number of items that were developed by OMIIEurope that are of interest to the OGSA-DAI project. OGSA DMI OGSA-DMI is a group that Dr Antonioletti had been involved in previously. He took over as co-chair from Michel Drescher, with Ravi Madduri from ANL, after Drescher found that he would not be able to continue in that position after other commitments made it difficult for him to dedicate the amount of time. On the whole, the DMI group has run smoothly and the specification recommendation went into public phase mode during this OGF. There is some commitment to do implementations from Globus, Unicore, Fujitsu and Microsoft to try to get the specification to full recommendation status. If successful, it could be adopted by other groups such as HPC Profile and JSDL to handle some of the file staging requirements. http://forge.gridforum.org/sf/go/doc15052?nav=1) Dr Antonioletti gave a presentation that: • Provided a brief overview of what had been done by the group. • Outlined some of the perceived gaps and future avenues of work. • Attempted to drum up support for someone to take over the chair role. Although there was some interest, and other areas were pointed out as worthy of investigation, no-one volunteered to take on this role. It looks as though the OGSA-D-WG work will go into hibernation until someone else wants to take the leadership role. DFDL-WG Dr Antonioletti attended the DFDL-WG session – Martin Westhead, one of the co-chairs, used to be at EPCC. It looks like the work is making progress despite the fact that they do not have many spare cycles. This work could be of interest, once it is completed, as a means of describing binary data which could be used in many different directions. SAGA+DAIS The SAGA+DAIS session was another attempt by SAGA to try and get some SAGA APIs for DAIS. Dr Antonioletti gave an overview presentation of what DAIS is as it has been confused with OGSA-DAI. There has been previous dialogue with the SAGA group – at OGF19 - but nothing came of it as there were no spare cycles to do anything. It did not seem apt at the time as there were no DAIS implementations, but it now appeared that that was beginning to change. An overview of SAGA was given and also of the current DAIS RDF work. The SAGA group was looking for use cases that can be used to inform an API. The DAIS-WG undertook to produce a number of these to pass on to the SAGA-WG to try and begin the process of generating an API. Standard API for Data Grid Management Systems BoF The SAGA group was out in force at the standard API for Data Grid Management Systems BoF. The SAGA group maintained that a lot of the proposed functionality may already be available through their APIs. Arun, who chaired the session, is going to talk to them and try to organise another BoF at a future OGF. Encyclopedia of Life The Encyclopedia of Life (www.eol.org) was presented by an ex-colleague, Jen Schopf. Her team’s current work was really high profile. Dr Antonioletti noted how they are very pragmatic about the technologies and standards they use. Page 61 3 GridNet2 Reports 3.3 Prof Malcolm Atkinson – NeSC Cloud Computing Dr Antonioletti found the cloud effort to be a little disturbing. From what he could make out, cloud computing was just a Grid but with a higher level of abstraction, which might potentially make it easier to use. This might be the next big thing, but it just appeared to be Grid computing in disguise. DAIS-WG As one of the co-chairs of the OGSA-DAI WG, Dr Antonioletti presented an overview of DAIS. The DAIS-WG has been in existence for a while now. Amy Krause, on behalf of the Ohio State University, presented an overview of the issues they ran into when trying to implement the WS-DAIX candidate recommendation. There were some tooling issues and also a lack of spare cycles. This was followed by Elias Theocharopoulos giving an overview of the OGSA-DAI issues that have been encountered when implementing the WS-DAIX candidate recommendation. This caused a lot of animated discussion which overran its time slot so the overviews of the RDF/RDF-S work had to be quickly glossed over and the session, as a whole, overran. It had to be drawn to a premature conclusion so people could attend the town hall meeting. The DAIS-WG was established by OGSA-DAI folks and at some level, it is still important to this project. There are a number of other middleware providers that are interested in implementing the WS-DAIR candidate specification and the RDF/RDF-S specialisations are very near completion. staff time, Mark Linesch’s time and much of the time of senior GFSG members. Nevertheless, Prof Atkinson endorsed the move on the basis of the argument that sponsors would not accept paying for both organisations. In hindsight, they had already stopped paying for EGA. The formal agreements, led by Jay Unger, as a new constitution were progressing well, but were dogged by legal and tax issues peculiar to the USA. Some time was spent reviewing the presentation and public image of GGF. The OGSA was seen to be a central coordinating framework – as intended – but GFSG was impatient about the rate of progress and definitive statements. This seemed to Prof Atkinson to be underestimating the difficulty of such coordinating architectures. Prof Atkinson reiterated his usual plea that SAGA was also a vital integrating influence which has high value in reducing the redundant, highly skilled software work that moves Grid applications. The review of the strengths of the areas and their groups received very little attention. This is unfortunate as they are the GGF’s principal product and raison d’être. It was good to have a number of side-bar discussions with other GFSG members on the state of Grid computing and technical progress. Many of these people Prof Atkinson would not meet anywhere else and would not be able to find out their views. 3.3.2 GGF16 and GFSG Data Management Workshop GFSG The Data Management workshop was interesting in giving a broad perspective of what is happening in the data space. Of particular note, though, was Paul Strong’s overview of how eBay works (sadly, the slides are not on-line). A number of impressive statistics were covered in his presentation. They do not depend on standards, which rely on the least common denominator, to achieve their impressive throughput and this does not always allow them to exploit vendor-specific optimisations. They seem to be keen to adopt standards at some future point but as with the EoL group, they take a very pragmatic approach. The GFSG meeting was well attended and followed on from the recent F2F in San Francisco so that the topics and distribution of time was much as described for the previous meetings. These same topics had dominated the weekly telcons for several months and it was very clear that this was diverting attention from standards and perhaps to community development. 3.3 Prof Malcolm Atkinson – NeSC Prof Malcolm Atkinson is Director of the e-Science Institute at the National e-Science Centre, Edinburgh, and is also the e-Science Envoy for the UK’s e-Science Core Programme. For a summary of Prof Atkinson’s GridNet2 activities, see 2.3, Prof Malcolm Atkinson – NeSC. 3.3.1 GFSG F2F Prof Atkinson attended the GFSG F2F meeting held in January 2006. The primary topic of GFSG was, as it had been for some time, the merger of GGF with the Enterprise Grid Alliance (EGA). This took the lion’s share of the day – discussing the new structure, new governance model and ways of presenting the new organisation. It seemed to Prof Atkinson that GGF was being significantly perturbed for very little gain. The issue was clearly absorbing a large proportion of the GGF Page 62 A review of the topics to be covered in the opening and town meeting had to manage the flow of publicity as well as sustaining the volunteer commitment. This is a difficult balancing act, particularly when also seeking to avoid ruffling any sponsors. GGF16 Prof Atkinson is Area Director for Data Area and the following covers his view of progress in the Data Area. It was not possible for him to attend all the WG sessions for the Data Area at GGF16. Being in Europe, the OGF sessions were well attended and it was pleasing to see plenty of work in the data area. The INFO-D -G seemed to be a little esoteric, more like a research group at times, but it was beginning to limit its scope and develop focus with planned applications. There was some continuation between the generic work on naming that is emerging as a requirement in OGSA and the particular melding of naming with file management in the GFS-WG. The DAIS-WG was making good progress on wrapping up its initial three proposals and developing a plan for interoperability testing. Steve Pickles was particularly helpful in advising the DAIS-WG on how to address this task. It became clear that this was pioneering new territory for GGF. It was frustrating not to have systematic work on distributed data management, including replication 3 GridNet2 Reports 3.3 Prof Malcolm Atkinson – NeSC and caching, but it had not been possible to get the relevant parties to work together. The OGSA-D-WG, led by Dave Berry, was, however, establishing a vision of requirements and a framework as to how these may be met. The GridFTP story was complete and the standard had become widely used, but the necessary experience report was not forthcoming to turn it from recommendation to a standard. The work on Grid storage management, which is the umbrella for standards shaping the Storage Resource manager (SRM) and is widely used by experimental particle physics, was very active but not particularly concerned with delivering a final document yet. Progress on DFDL was slow, which was a pity as it is a very important approach to efficient, interoperable data access. The generic requirement of a data movement request abstraction was beginning to be recognised by both the GSM and GridFTP groups. Prof Atkinson was also acting as initiator for the work on training and education. David Fergusson and Prof Atkinson had organised a workshop and BoF on Grid education, partially motivated by the imminent start of the FP6 ICEAGE project. The workshop and BoF were very well attended with a lot of energy and promise of commitment. David Fergusson and Prof Atkinson therefore resolved to propose the formation of a Training and Education group. Prof Atkinson has managed to get this as one of the named goals of the new organisation. Prof Atkinson used all of the opportunities available, including an announcement at the Town Hall meeting, to advertise ISSGC06, which is endorsed by GGF who provide administrative help through Ann Colins. Overall, Prof Aktkinson found this to be a successful and effective GGF meeting with much going on throughout the whole day on each day of the meeting that was driving forward the Grid agenda on many fronts. It was a little surprising to experience a dust storm from the Sahara while in Athens. 3.3.3 GGF17 GGF17 and the integrated GFSG meeting were held in Tokyo, and the public transport adventure of getting to the GFSG proved surprisingly easy despite the profusion of lines and platforms in this immense and complicated network; thanks to punctual schedules, clear English notices and, above all, Satoshi Matsuoko’s careful instructions and meticulous preparation. He was a wonderful host throughout the event. The GFSG was, once again, dominated by the practicalities of the transition to the merged OGF. There is a considerable challenge keeping arrangements developing without going public. The Enterprise community still seems to be misguided in their belief that GGF, a standards organisation, can do market building. This is inheriting the worst features of EGA and it does not deliver a programme that will convince anybody, as the unconvinced would not attend GGF/OGF meetings – another misguided distraction. Worse, they have the notion that they can gather requirements that the standards groups will implement. There are two intractable problems with this, as follows: 1. The discussions are invariably at a generic level which fails to capture the detail needed for standards and usually concerns the composed functions that depend on the integration of many standards. 2. The work in standards working groups has usually started with collection, collation and analysis of requirements, and they then have a momentum that will not be diverted, and should not be, until the standard is produced. These are symptomatic of challenges for GGF (or any similar SDO), as follows. 1. It is very hard to recruit and sustain the efforts of technical experts relevant to a standard. 2. It is very hard to steer volunteers to do anything other than the work which they originally came to GGF to undertake. Usually, this work is for one or two specific standards that warrant the allocation of their time in the judgement of themselves and their employers. 3. Standards require development of consensus, documents that describe accurately that consensus, and reports from independent implementations that show the standards in use. Those not engaged in standards development invariably grow impatient and denigrate the effort because they do not appreciate the need for this elapsed time to cover the whole process. So, for it to be a benefit, the formation of OGF must address these issues. In particular, it must generate effort from industry that includes experts working on standards. If industry requires standards, it must be prepared to invest in them. Though there are promising signs that Microsoft will invest in the OGSA execution model HPC profile, there is no general commitment. Paul Strang is clearly a very helpful, powerful intellect with commitment and deep experience, but there is a very real dearth of prospects in the data area despite many data companies being in the EGA. Switching to sessions, the DAIS-WG meetings and the OGSA-DAI workshop were a great success as the very strong local community, led by Isao Kojima, was very active. In particular, they had well developed experience with RDF extensions to OGSA-DAI in order to hold biomedical semantic data. This led to a commitment to develop an RDF version of the WSDAI specification. Prof Atkinson was also delighted to hear that OGSA-DAI is a key component of NAREGI and in production use for their Grid information system. Looking at the Data Area as a whole, Prof Atkinson’s view as Area Director was that this was a most productive meeting. Though the number of people present was reduced, the number of active and effective standards workers was not – and they were not distracted by explaining things to people who only show up once. (There does seem to be some case for constraining participation, at least in formulating and drafting standards, to those who have read the WG’s documents and informed themselves of the relevant background.) The INFO-D-WG was particularly productive after a pre-meeting. The BYTE-IO-WG was making spectacular progress on requirements sparked by DAIS and OGSA-DAI. Professor Atkinson stated that Neil Chue Hong and his colleagues from Virginia were to be congratulated. Page 63 3 GridNet2 Reports 3.3 Prof Malcolm Atkinson – NeSC Overall, all of the standards areas seemed very productive – possibly a demonstration of the ‘mythical man month’ effects. The talk on OGSA by Hiro Hishimoto was excellent and made it exceptionally clear why an overwhelming framework in which to build standards is reliable. Though it had rapidly advancing profiles (specific realisations), there was still a lack of realism in the impatience about progress. David Snelling clearly understood and balanced the issues, but Mark Linesch was keen to pick just one ‘silver bullet’ to demonstrate GGF’s value. Prof Atkinson still felt much more should be made of APIs, particularly the work of the SAGA-WG, as immediate gains can be made in productivity and portability through stable APIs. The putative group on Education and Training held another BoF, with much energy and enthusiasm. The BoF (mainly David Fergusson, Kathryn Cassidy and Prof Atkinson) thrashed out a proposed programme of work on charter, which may prove to be a bit ambitious. The contributions from Rüdiger Berlich were significant. Reagan Moore gave an interesting talk on his plans for iRODs, but it connected very little with standards. Instead, the talk explored a re-implemented SRB with rules/computation that can be triggered by data actions. Microsoft demonstrated a multi-site conference system. It looked easy to install and use but did not interwork with Access Grid. Furthermore, it would depend on the same difficult to tune Gentner echo cancelling and poorly-supported multicast standards. A report on the meeting could not be concluded without a comment on the wonderful building in which the event was held: a remarkable open inverted hull shape on an enormous scale, with shelves of rooms around the sides connected by inner airy walkways and dramatic transverse level-changing bridges. The GGF meeting only used a small fraction of the space, which had its own metro station in the basement. 3.3.4 GGF 18 This was the last GGF/OGF where Prof Atkinson had to arrive on the Saturday for Sunday’s GFSG as he had decided that as he is on the OGF Board, he could not also be on the GFSG. Prof Atkinson reviewed the experience of being on GFSG, as follows: Initially, it was a very valuable activity but timeconsuming, involving the following: • Three one-day F2F meetings (held before each GGF) • Attending all of each GGF, with side-bar meetings, running approximately from 7.00am to 9.00pm each day • A two-day F2F in January • A half-day F2F at Super Computing • Weekly two-hour telcons • Reading reports • Communicating with group chairs Page 64 Membership of the GFSG had a high value due to participation in the following activities: • Providing interesting networking with field leaders, as well as other Area Directors, the GGF office team and the GGF leaders • Supporting decisions on managing groups • Imposing quality controls on the GGF publications • Planning the GGF events • Addressing sustainability and publicity The dominance of this merger and the business it generated greatly reduced the time invested in standards, even though the time, telcon calls and F2F meetings were officially partitioned to preserve the division of attention. Extra calls and F2Fmeetings were undertaken to address pressing standards issues, such as priorities, integration and gap analysis, and document processing. After discounting this, there were still two major challenges, which remained insuperable. 1. Although labelled as a steering group, it was very hard to steer. a. The standards (and other groups) worked on what they cared about (or their employers wanted) and it was hard to get anything else done. Group steering was mainly limited to persuasion to achieve minor changes or termination if the group became inactive. Blocking publication of a report made a group reconsider, but this tactic had to be used sparingly. b. Where requirements were recognised, it was not always possible to persuade people to start a group to address them. c. After many years of trying to influence the form of GGF/OGF events, it was clear that they were not shaped by GFSG and controlled by their programme committee. The exigencies of satisfying sponsors, the diverting pull of publicity, the varying ideas about enterprise and e-Science, and the president’s need to commit during crucial meetings combined to overwhelm any principled plan. 2. The continuing challenge of obtaining sufficient income often led the organisation along contingent paths or meant that the resources to achieve the goals were not available. The office staff, led by Steve Crumb, worked wonders but they are not magicians. Prof Atkinson recommended that UK e-Science had one or two active members of GFSG – for example Dave De Roure does a vital job at the interface with W3C that is useful to UK e-Science – but suggested that, in the future, the UK should have only a modest presence in the GFSG. The UK had had a substantial proportion of the Area Director chairs in the past. Prof Atkinson was sorry to relinquish the role of Data Area Director and noted that he would miss the interactions with David Martin, the co-director of the Data Area, and the work with the individual working groups. 3 GridNet2 Reports 3.3 Prof Malcolm Atkinson – NeSC The atmosphere at the event was frenetic because of the juxtaposition of GridWorld. There was no evidence of benefits to GGF and GridWorld being colocated, and the WG sessions were depleted of effort compared with their normal attendance and energy. The data area continued to push effectively on many fronts, including the following: • DAIS, where there were documents in comment and interoperability testing, a progressive handover of chairs (Dave Pearson and Norman Paton did an excellent job). The RDF story was developing with significant input from Isao Kojima and the ONTOGRID group in Spain. • The GFS-WG was progressing well, but there was emerging contention over the issues of standards for naming, as the OGSA-WG had a vision of a generic naming standard and the GFS-WG was trying to accommodate the existing behaviour of file systems. This was eventually resolved. • The work on storage systems in the GSM-WG, essentially a standard for SRM, was progressing slowly on paper but well in practice with >5 implementation efforts under way. It was unclear how this related to other storage activities, such as SRM and SNIA standards, but it related well to GridFTP. • GridFTP was productive; that is, it yielded a standard which is in daily high-relevance production use, including by the above SRM stores. This was a success for GGF, which was well hidden because the Grid FTP community had not written an experience report to transform the ‘standard’ from a ‘standard proposal’ to a standard. • The above two communities combined in a group to define a Data Movement Interface for requesting a bulk movement of data, something which is much needed. They were focusing on file movement. The OGSA-D-WG had also identified the need for DMI. • There was not much sign of DFDL progress, which is a great pity. • The Byte-IO-WG had demonstrable success and was still moving forward quickly. • There was still no prospect of forming an effective group to look at distributed data replication, though this was clearly important and there were several operational implementations. A welcome first at this GGF was the active engagement of the Open Geospatial Consortium (OGC) standards group. This is very significant for the standards area as about 80% of data has some geospatial element. The OGSA-DAI team, under the aegis of OMII-UK, had a very successful tutorial workshop and user group meeting. The Education and Training Community Group (ETCG), under the e-Science OGF banner, received its charter approval just in time for GGF18. The group held its first two sessions, setting out its programme of work and organising its several co-chairs. There was strong input from Wolfgang Gentsch of D-Grid and other colleagues from Karlsruhe, and significant interest from the USA for the first session. The second session was poorly attended, a common feature of this GGF. Most of the detailed planning for ET-CG had been done by David Fergusson and Prof Atkinson, and they chaired the sessions. Their thanks were given to Joel Replogle for setting up ETCG in about 24 hours, and to Kathryn Cassidy for a burst of secretarial energy recording our meetings and populating our ‘wiki’. The formation and announcement of OGF was, by this stage, a non-event for GFSG members as the subject had been discussed ad nauseam for over a year. The big question was whether it would bring in industrial technical commitment to work on standards and experience-led best practice. Paul Strong was an immense asset, gained through the merger. This was very evident at the Board, as well as when he spoke in plenaries and session. Prof Atkinson did not have significant other evidence of gain in this area, but he had a limited view and there was time for other positive gain to be shown. The other big question was whether it will help sustainability and income. The software developer track proposal by Charlie Catlett was a good one for gaining engagement and community strength. The OGSA was progressing well, particularly on the Basic Execution Service (BES) front, and there was growing interest from Microsoft in developing a BES HPC profile. Overall, this event was a mixed performance, some very good sessions but others where logistics and rival activities limited expert engagement severely. 3.3.5 OGF19 Prof Atkinson found this to be an excellent OGF because it was in a very good location in a spacious building, which was in quiet surroundings, and was somewhat off the beaten track. The facilities were excellent and there was no distracting razzmatazz. Much of Prof Atkinson’s time was spent talking in side-bar discussions with people from the software development community because this was the first occasion on which there had been a software development track, an idea that originated from Charlie Catlett in the Board meeting, and turned out to work well. The other main activity in which Prof Atkinson was extensively engaged was the whole series of meetings on the Grid Education and Training activity. This was very effective because a good group of people attended the meetings, including people who came especially for one day from places like Indiana. The programme of developing the Education & Training documents picked up momentum at least in the area of t-Infrastructure, the sustained infrastructure to support training or education, shared standards for accreditation of training, and shared standards for IPR. The challenge of developing a view on policy did not develop quite as far as Prof Atkinson had hoped. Once again, the DAIS-WG progressed well. The standards were now well advanced and activity on the RDF was beginning to develop. Overall the OGF meeting was a great success, doing the sorts of things that Prof Atkinson would like the meeting to do. Page 65 3 GridNet2 Reports 3.3 Prof Malcolm Atkinson – NeSC 3.3.6 OGF20 A great deal of work had been done by the UK community to prepare this OGF which was held in Manchester, UK. There was a large exhibition of work put on by the e-Science community in the UK and the same space was used by EGEE. This was a very large conference because the EGEE meeting was combined with the OGF meeting, and this was very successful, with well over 900 people attending. Much of the local effort had been led by Dave Berry from NeSC and Stephen Pickles from NWeS. There was also a strong programme of enterprise activities developed by the GCN! consortium, Dave Berry and Ian Osborne. The programme contained many interesting talks and Prof Atkinson particularly noted the value of the talk by Paul Strong from eBay on the complex and challenging problems of running a Grid which keeps all of the data properly replicated, all of the software properly deployed and upgraded on the system that supports all of the eBay transactions, which is many thousands of transactions per hour, and this has to be sustained all of the time because many businesses and individuals depend on it. It was a good example of very challenging Grid technology. The problem that Paul has is that very little of it is as yet built on either standards or de facto industry standards. The work of the ET-CG developed well. There was a significant number of people at the meetings and the group invested quite a lot of time on discussing and developing plans for consistent training policy. They also developed views on managing IPR for educational training policies. This was very important as they have to share material in such a rapidly changing domain. New material is required often but exercises with example data and example software are very labour-intensive to build and so it really is worth sharing such exercises between institutions. However, the institutional ownership issues have to be understood by the academics who prepared the material. We talked at some length about cooperation on the provision of t-Infrastructure, support training or education, and again, there are big advantages for sharing, not least coping with the very peaky load of training sessions, but policies are needed which allow that sharing to be fair. The group made progress with ideas regarding professional Grid qualifications. The work in the DAIS-WG also made progress, again particularly in the RDF area. The work on OGSA naming finally produced a result and it was nice to see this. There was some resolution of the relationships with Grid file systems as well and there were good demonstrations of that work by the team led by Andrew Grimshaw. Overall there was good progress on reference models and detailed definitions that support interoperability and sharing and management of computational systems but still there seemed to be too little work on the management, sharing, distribution and replication and so on of the large volumes of data, for which Prof Atkinson thought Grids were very significant, and there was virtually no work on how to get connections to legacy data repositories. The one community where there was some interesting collaboration going on in this area was the Open Geospatial Consortium interacting with OGF and that was now a significant development. Again some of this was led by people in Edinburgh, from EDINA. Page 66 3.3.7 OGF21 This is an OGF which Prof Atkinson had to attend as briefly as possible because of UK commitments and the need to minimise jetlag as a result. It was good to meet a number of former UK e-Science community members and to note that they were giving maximum benefit of their experience and the UK investment in e-Science to a company which likes to extract as much money from us as possible. Prof Atkinson wondered what could be done to try and encourage better use of the advanced knowledge and skills within the UK. Prof Atkinson’s primary reason for being at this OGF, apart from attending formal meetings, board meetings and so on, was to push forward the ET-CG activities. The overall discussion of policy and Grid education still tended to involve many issues on terminology but, overall, there was consensus that investment was necessary, that sharing was necessary, that sharing ideas on curricula was necessary, and there was a endorsement towards the development of common practical material. The framework for professional Grid certification proved quite challenging. The proposers of the current form of the document from Germany had a very strong view that this should be modelled on the sort of professional certification that is undertaken by Cisco and Microsoft or the Linux community for authorisation of people to work on their technology. However, the Grid is much broader and wider than that and so it is necessary to look for a more flexible and more general process. Their view was that the OGF could take on that role. That was not possible because the OGF does not have the academic credentials or the legal framework and certainly most countries expect to have that kind of accreditation achieved via the professional bodies or accreditation bodies within their countries, endorsed by their Education Ministry. So, finding an international model was still something of a challenge. The discussions on managing IPR for educational repositories were not progressing as fast as Prof Atkinson would have liked, mainly because no one from the USA (a major potential contributor), or from Asia, was coming forward and being prepared to put forward their views on how they would manage IPR. The proposal to build on the Creative Commons was a good one but the difficulty is that IPR legislation and standard practices differ greatly in different countries and, in many cases, the academics who are creating the IPR are not fully aware of their employer’s attitude to sharing it in a common repository, that is, they are apt to sign (tick) releases when they do not have the authority to do so – a notion of informed consent is needed. Implementation of policies for cooperating on t-Infrastructure was another topic which was progressing well but maybe not as fast as Prof Atkinson would like. The difficulty here was not so much interoperation as cohabitation. The problem was to get the various models of Grid software and the example applications that are used for education to be able to run in the same context so that lecturers and other educators do not have to build or access multiple contexts in order to demonstrate the multiple concepts in this domain. The work was continuing with most of the effort of writing being undertaken by the team from Catania. 3 GridNet2 Reports 3.4 Prof Mark Baker – Reading The conference showed two very strong successes on interoperation which Prof Atkinson thought was a sign of the improvement and maturing of OGF. The OGSA ByteIO interoperability first demonstrated two independent implementations of ByteIO based on the proposed standard and this, of course, was the requirement to take a proposed standard recommendation to a full standard. That work was led by people at EPCC in Edinburgh and it was very good to see it coming to a success. 3.5 The other interoperability success which was quite spectacular was the work on the HPCP. That had been developed in a very short time and had built on the JSDL standard which was developed and proposed by the UK and led by people from Imperial College London. It was good to see this progress. 3.5.1 Overall, it was a successful if somewhat brief, for Prof Atkinson, meeting in Seattle. 3.4 Prof Mark Baker – Reading Prof Mark Baker is Research Professor of Computer Science in the School of Systems Engineering at the University of Reading. For a summary of Prof Baker’s GridNet2 activities, see section 2.4, Prof Mark Baker – Reading. 3.4.1 ICEAGE - Curricula Development Workshop The GridNet2 award funded Prof Baker’s attendance at the ICEAGE Project Curricula Development Workshop for e-Science Education that was held in Brussels on the 14- 15 February 2008. The ICEAGE Project Curricula Development Workshop for e-Science Education had the goal of gathering together educators interested in discussing and laying out a syllabus that could be used to educate and train both undergraduate and Masters students that wish to pursue courses related to e-Science. The workshop was attended by a fairly diverse group of people ranging from those involved in computer science education to individuals working in particular e-Science fields, as well as those interested in general education. On the first day of the workshop, the group discussed and debated the prerequisites and educational goals for an undergraduate level e-Science course that was made up of three stages, including e-Working, as well as basic and advanced research methods. On the second day, the prerequisites and educational goals for a Masters course in e-Science were outlined and discussed. Overall, the discussion and debate that took place over the duration of the workshop provided a valuable contribution to the progress of defining a curriculum for undergraduates, as well as for postgraduates. In addition, the workshop allowed the people attending to the workshop to meet, socialise, and exchange views on e-Science in general. Funding from GridNet2 was essential to enable Prof Baker to attend and contribute to the Project Curricula Development Workshop for e-Science Education. The outcomes of this work will be of great use to the e-Science community, and will underpin future educational developments for e-Science and e-Research. Dr Dave Berry – NeSC Dr Dave Berry is Deputy Director of the e-Science Institute at the National e-Science Centre, University of Edinburgh. For a summary of Dr Berry’s GridNet2 activities, see section 2.5, Dr Dave Berry – NeSC. OGSA F2F – Jan 06 Dr Berry attended this meeting of the OGSA-WG in his role as co-chair of the OGSA-D-WG and as a member of the OGSA-WG. One session of the meeting was devoted to the OGSA Data Architecture, discussing in detail how it fitted with the main OGSA architecture and with other OGSA activities such as ByteIO (which was work spawned by the OGSA-DWG and which eventually published a specification in Jan 2007), WS-Naming, RNS (to which the group recommended several changes to make it better fit with OGSA) and EMS. The OGSA-WG also worked on version 1.5 of the OGSA Document. This document, of which Dr Berrry was a co-author, was finally published in September 2006. 3.5.2 GGF16 Dr Berry attended GGF16 in his role as co-chair of the OGSA-D-WG. The OGSA-D-WG held two sessions on the current state of their two documents. These were useful in getting feedback from the community about the current state of these documents. Dr Berry and the OGSA-D-WG also spent time outside sessions incorporating this feedback into the document. The group’s work on the OGSA Data Architecture led to the organisation of a BoF on Data Movement Interfaces (DMI). This was held at GGF16 and led to the creation of a working group to develop a standard for a DMI interface. This is supported by several leading e-Science Grids, including CERN/EGEE. The joint session with SNIA covered some possibilities for co-operation. There is clearly some overlap of interest as management of storage underpins data Grids. Dr Berry attended the following sessions: • Overview of the OGSA Data Architecture (OGSA-D-WG) (co-chair) • OGSA-D Working Session (OGSA-D-WG) (cochair) • GIN • GFS-WG • Grid Architecture Experts Workshop • Data Movement Interface Standardisation (DMIS BoF) • SNIA Joint Session • DAIS Working Group Session (DAIS-WG) • Data Area Meeting Page 67 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent 3.5.3 GGF18 Dr Berry attended GGF18 in his role as co-chair of the OGSA-D-WG. The OGSA-D-WG held two sessions to review the current state of their two documents. These were useful in getting feedback from the community about the current state of these documents. The group also spent time outside sessions incorporating this feedback into the document. In addition, Dr Berry attended a range of other sessions in order to keep himself up to date with developments in technology and GGF standards. Of particular interest was the session on integrating OGSA and the EGA reference model, which was aimed at ensuring that models of Grid from enterprise and e-Science would work together, and identity management, which reviewed the state of the art in Grid security. Dr Berry also attended a number of enterpriseoriented workshops, partly in his capacity as Technology Lead of the Grid Computing Now! Knowledge Transfer Network. These workshops enabled him to keep in touch with the growing uptake of Grid in industry and, of more importance, to track what they see as the most important outstanding issues facing the uptake of Grid technologies. The keynote presentation by Paul Strong of Ebay was particularly interesting, exploring the issues of managing enterprise infrastructures on a huge scale. Sessions attended: Architecture document. They resolved many outstanding issues, clarified others, and revised the overall structure to improve the readability of this document. They also worked on the companion document on Scenarios for the Data Architecture. In particular, they led a joint session with people working on OGSA EMS scenarios. This led to revisions on both sides to ensure that the two strands of work would work together. In addition, they reached out to other working groups in the data area, to co-ordinate their work and that of the OGSA Data Architecture. Sessions attended: • OGSA Data Document Review (OGSA-D-WG) (co-chair) • EMS/Data Scenarios (OGSA-D-WG) (co-chair) • OGSA Workflow (OGSA-WG) • OGSA ByteIO WG Session (BYTEIO-WG) • Standards All Hands: Integrating the Work of OGF (2 sessions) • Needs and Requirements for Quality of Service in the Grid (BoF) (Quality of Service BoF) • The Storage Resource Management Standard Proposal - the Core Version (GSM-WG) The OGSA Data Architecture document was finally published in December 2007, with Dr Berry as one of the co-authors. • Grid Computing Now! The UK Experience in the adoption of Grid Computing Technologies (cochair) • OGSA Data Architecture Services (OGSA-D-WG) (co-chair) • The OGSA Data Architecture Scenarios (OGSAD-WG) (co-chair) • Enterprise Grid Requirements • Mini Symposium - Development Tools for GT4 Service Programming • OGSA and EGA Reference Model (OGSA-WG) 3.6.1 • Pharmaceutical Grid Requirements OGSA-Authz-WG Meeting • Enterprise & Standards Summit: Session 1 - Requirements Rollup, and Session 2 - Requirements Prioritization This was the first meeting that Prof Chadwick took over as a joint chair with Von Welch. • Topics in Identity Management 3.5.4 OGSA WG Telcons – Aug 06 Dr Berry contributed to the OGSA working group, primarily to co-ordinate work on the OGSA Data Architecture with the overall developments of OGSA. This also allowed him to keep in touch with developments in OGSA. 3.5.5 OGF19 This was a very productive meeting for Dr Berry. In addition to the formal sessions, Dr Berry spent a great deal of time working with his co-chair Allen Luniewski on the draft of the OGSA Data Page 68 3.6 Prof David Chadwick – Kent Prof Chadwick is Professor of Information Systems Science at the University of Kent. For a summary of Prof Chadwick’s GridNet2 activities, see section 2.6, Prof David Chadwick – Kent. GGF14 The status of the current deliverables was briefly discussed – the Attribute document was in hands of GGF Editor, and the SAML profile document would be at end of WG last call at end of this week. Attention was now focusing on the next set of deliverables. Prof Chadwick gave a presentation about authorisation architectures in a multi-domain environment (the slide show is on Gridforge site at https://forge.gridforum.org/sf/go/doc13703?nav=1). Prof Chadwick’s presentation considered which components are needed and how the target Source of Authority can remain in control of the policy for access to its resources. The concept of a Credential Validation Service was introduced to the group. Frank Siebenlist gave a similar presentation but from a different viewpoint. This considered asking remote 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent domains (AAs) if they could help in the credential validation. Dane raised the issue that if one network round-trip to each security service is required, then this would cause a major performance problem. There was quite a bit of discussion about the correct way forward. No agreement was reached, but it was recognised that the problem space needed to be constrained for the next set of standards, since it would not be possible to standardise everything in the next round. The meeting then discussed the Charter Revision to cover the next round of standardisation effort. The first two paragraphs of the existing charter were all right for the revised WG but the third paragraph onwards needed revising since it talkedabout a twophased approach. It was recognised that more than a two-phased approach was needed, since OGSAAuthz-WG was already in phase 2. It was agreed that this WG would be an authorisation architecture group for OGSA even though its output would be aimed at a wider audience than simply OGSA. It was also agreed to retain the OGSA-Authz name for the group. The tentative set of output documents were agreed upon, as follows: 1. A scenario document which can be part of the architecture document 2. Version 2 of the PEP-PDP protocol document 3. Version 1 of the PEP-CVS protocol Implications for UK e-Science It was still very early days for standardising a replacement protocol for the OGSA Authz SAML profile. The latter was known to be deficient (not least from experiments carried out by Richard Sinnott at Glasgow) but there was no consensus yet as to what the replacement should be. 3.6.2 GGF15 OGSA-Authz-WG Meeting Prior to the meeting, Von Welch had prepared and circulated an outline scheme for combing Shibboleth, GT4, VOMS, and PERMIS. This showed the interconnection points and the standard interfaces that would be needed. Prof Chadwick had hoped that this would be a major focus of the meeting, but this did not turn out to be the case. Instead, a lot of time was spent on discussing the development of XACML v3 by OASIS, which is expected to be completed by the end of the year, and how this might be a solution to many of the authorisation problems. This tended to divert attention away from spending time working on the next generation of authorisation protocols, which had been agreed at the last meeting, and instead treading water waiting for XACMLv3. Consequently, Prof Chadwick found this meeting to be somewhat disappointing, and not to move things forward in any way. The previous meeting had recognised the need to progress from the current SAML Authz protocol to something with more functionality. Prof Chadwick had expected this meeting would make significant progress towards the next generation protocol(s). However, this was not the case and, in Prof Chadwick’s opinion, no further progress was made during this meeting. Implications for UK e-Science It was possible that XACMLv3 would provide the extra needed functionality, for example, in the area of delegation of authority, but this would not be clear until it is delivered. In Prof Chadwick’s opinion, this should not be a reason to postpone developments altogether. As Bob Cowles pointed out at the meeting, there was considerable value in an iterative approach to standards and releases, rather than a "shoot-forperfect" approach. Thus UK e-Science should migrate from where it currently was to something that is better, based on what existed at the time, rather than waiting for the next version of XACML, since there is no certainty what this would contain or when it would be delivered. Leveraging Site Infrastructures for Multi-Site Grids Workshop This was a very varied workshop with nine presentations showing the wide span of activity in this area. There were talks about Shibboleth, GridShib, Condor-Shibboleth, Pub cookie-MyProxy, Signet-Grouper. Prof Chadwick gave a presentation about dynamic delegation of authority between sites, which was being piloted between Glasgow and Edinburgh as part of the JISC DyVOSE project. Whilst the workshop itself was a good way of disseminating information, and learning about these different projects, it also highlighted the complexity and diversity that still exists in the security world, and that fact that there were still many different ways of achieving the same functional requirements of authentication and authorisation. Implications for UK e-Science Whilst UK e-Science was still in the phase of “let a thousand flowers bloom”, there would eventually need to be some shake up, and decisions to back some horses instead of others, otherwise interworking between sites would be very difficult. Shibboleth appeared to be a winner, and integrating Shibboleth with Globus Toolkit and OMII should be a major focus for e-Science. 3.6.3 GGF16 As usual, a major benefit from attending GGF meetings was the unprecedented ability to network with a large group of researchers who were closely involved with similar research issues. This meeting was no exception, and Prof Chadwick managed to talk to many people about forming a Network of Excellence in Grid security. Should such a NoE be formed, this would, of course, have major benefits to UK e-Science. OGSA-Authz-WG Meeting Prof Chadwick co-chaired the OGSA-Authz-WG meeting with Von Welch. The meeting initially discussed the revised charter that Prof Chadwick had circulated prior to the meeting. A number of additional issues and deliverables were added to the revised charter, and Prof Chadwick later distributed an updated version to the list. Prof Chadwick then presented his vision for how authorisation in VOs should evolve from their current status, with the additional of an attribute credential validation service (CVS). The functionality of this module was to ensure that the attribute credentials were valid prior to making an authorisation decision, and extract the valid attributes. Invalid credentials would be Page 69 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent discarded. There was wide agreement in the group that this functionality was needed, and that current PDPs, such as XACML, expected to receive already validated attributes. Standardisation of a protocol for communication between the PEP and the CVS was a proposed deliverable in the revised charter. Nate Klingenstein from Internet 2 gave a short talk at the end of the meeting about the problems of authorisation using the attributes from multiple attribute authorities (AAs). The problem arises because the user was typically known with different identities in each AA. He suggested two models. The model that was easiest to implement was where attribute credentials issued by multiple remote Attribute Authorities (AAs) were validated by the Identity Provider’s local AA, their signatures were stripped off, and the attributes combined with the local ones and transferred to the Service Provider (SP) as one combined attribute credential. The SP then only needed to validate one credential. However, this model was weak from a trust perspective, since the local AA was asserting that a user has attributes which it is not an authoritative source for. The more difficult model to implement was one in which the SP actually retrieved the attribute credentials from multiple AAs. How did the SP know the different identities of the user that were used by each AA? After the meeting, Prof Chadwick thought about this problem and, on the following day, wrote an outline solution to the problem. He expected this outline to be submitted to the Internet2 consortium for review, and then finalised for presentation at a conference later in 2006. Shibboleth Globus Toolkit BoF This BoF was chaired by Von Welch, and was split over two days. The first day comprised 7 presentations from researchers who are working on different aspects of merging these two technologies together. Prof Chadwick gave a short talk about his GridShibPERMIS and GT-PERMIS projects. Participants from the University of Manchester gave a short talk about the SHEBANGS project. The second day comprised a discussion of the various issues and drawing up a list of actions that should be undertaken. This work was of importance to the UK, since JISC is supporting the rollout of Shibboleth to university computing services, whilst the NGS is based on Globus Toolkit. Clearly the merging of Shibboleth and GT technologies would be beneficial to UK e-Science. 3.6.4 GGF17 OGSA-Authz-WG Prof Chadwick chaired the WG meeting alone as unfortunately Von Welch could not make this meeting. Significant progress had been made since the last GGF meeting. The Charter had been revised and two IDs had been published: the PDP and CVS protocol specifications, both written by Prof Chadwick and his team at Kent. All three documents were discussed in two long telephone conferences in the weeks prior to the meeting. The first phone conference suggested that the charter should be slimmed down dramatically and should only fit the work which the OGSA-Authz-WG could do, that is, produce documents that for which the group had editors. This proposal was confirmed at the meeting, and so Prof Chadwick agreed to produce a revised charter. The second phone conference concentrated on the CVS protocol specification and the use of XACMLv3. Frank Siebenlist proposed that XACMLv3 could be used for both the PDP and the CVS protocols. Prof Chadwick agreed that it could be used for the first, providing a stable version is published by OASIS, but doubted if it could be used for the second. This was discussed further at the meeting and it was agreed that Prof Chadwick should draft a list of requirements for a CVS. Two members of the OGSAAuthz-WG volunteered to analyse the requirements to see if XACMLv3 was able to satisfy them. The relevance of this work to UK e-Science was that if no agreements could be reached on architectures and protocols for authorisation infrastructures, then the provision of authorisation components would remain fractured and disjointed. UK e-scientists would either have to make do with what was already provided, or patch in their own additional essential bits and pieces to give them the added functionality they required, without enjoying the benefits of plug and play, and the ability to select best of breed that standardised interfaces would provide. This sticking plaster approach applied in many application scenarios, with many components, such as LCAS, gridmap-files, EDG Java Security, GridSite, G-PBox, PERMIS, XACML and GACL, all being used in an ad-hoc and uncoordinated way. The duplication and wasted effort that this caused was enormous, with seemingly every application developer inventing their own approach to authorisation provision. Various presentations were given, the most noteworthy being from Blair Dillaway of Microsoft who gave his vision of future Grid security, and the numerous issues that need to be addressed this state of Grid security is achieved. One of the disappointing features of the current GGF OGSA-Authz-WG meetings was that some authorisation infrastructure developers, such as the VOMS team, did not actively participate in the meetings. Prof Chadwick wonders if they would actually adopt the protocol specifications that would eventually be produced by this WG. This had to be a cause for concern for UK e-Science. GGF Chairs Update Astro-RG Work Shop This meeting updated the chairs of the GGF groups about the new facilities that would soon become available on GridForge. There was also an update on other issues of importance to the GGF, such as the upcoming merger with EGA. The subtitle of this day long workshop was Building a Global Virtual Observatory for Astronomy - Grid Standards and their Relevance in Use. Chadwick was invited to speak about the work of the OGSA-AuthzWG. Security Area Group Meeting Prof Chadwick found the group to be very receptive to the work within OGSA-Authz and desirous of a Page 70 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent standard interface for plugging in different policy-based authorisation infrastructures. The Astro-RG was already tackling problems in this area. After Prof Chadwick had finished his presentation, the group said that they would feed back their requirements into the OGSA-Authz-WG, which would be very beneficial to it. 3.6.5 GGF18 Shibboleth for Grids This comprised four sessions, two on both Monday and Tuesday, which were devoted to updates about various projects around the globe that are integrating Shibboleth with Grid technologies. There were presentations from the following people: but not the same as XACML). The language provides safety properties via simple syntactic checks. MS wanted to define complete policies that specify trust, delegation and authorisation. They use standard XML parsers and XML DigSigs to protect their policies. Principals are identified by their public keys rather than their names. Resource hierarchies are supported. Attributes are name value pairs (this is pretty standard now). They use regular expressions for matching variables. All-in-all the system has some good features and MS has tried to pick what they see as the best features that existing authorisation PDPs/policies provide. Jef Tan from Monash University talked about connecting across firewalls for the Grid. He used a combination of SOCKS and SSH to get past university firewalls. • Erik Vullings, via Skype from Australia, who talked about the MAMS system and tools such as Autograph that control attribute release policies • Christoph Witzig, of SWITCH, who is working within EGEE to add Shibboleth authentication as an alternative to X.509 PKI authentication • Von Welch, who gave an update about the GridShib project • David Spence, from CCLRC, who gave an update about the ShibGrid project • Mike Jones, whose talk about the SHEBANGS project raised more unresolved issues than answers, and who also gave a summary about GridSite and Shibboleth • Richard Sinnott, who spoke about the various projects at NeSC Glasgow The minutes of the OGSA-Authz-WG meeting are available at: https://forge.gridforum.org/sf/go/doc13860?nav=1. • Prof Chadwick, who described the latest user friendly tools for setting PERMIS policies for Shibboleth and Grid resources; 3.6.6 • Nate Klingenstein, who gave a comprehensive talk about SAMLv2 and Shibbolethv2 • SAGA security discussions • Alan Sill, who spoke about combining VOMS and the registration service VOMRS with Shibboleth • Security Area meeting • • Federated Identity workshop Tom Scavo, who talked about X.509 and SAML bindings • LOA BoF • CAOPS-WG • OGSA Authn Charter Clearly there was a lot of activity in this area and there were many similarities in the approaches, which involved having an online pseudo-CA that issued short-lived certificates for use on the Grid, based on Shibboleth authentication of the user. It would be good if one common model for this could be standardised. Globus Security for the Real World This session described not only the current security mechanisms in GT4, but also the future mechanisms that Globus was currently working on, such as improved authorisation. Security Talks Blair Dillaway from Microsoft gave a talk on access control research that MS has been doing. SecPAL is a declarative, logic-based, security policy language. It supports the distributed creation of policies. Permissions are monotonic so the results are always predictable (this was the same as the PERMIS model, OGSA-Authz-WG meeting The meeting was very productive, and had been preceded by a teleconference for those WG members who could not attend GGF18 in Washington. The revised charter for the WG was finally agreed, with only minor amendments being suggested to the version that had been posted two months prior to the meeting. The CVS Requirements document had some more features recommended for inclusion in the next version. The Authorisation Functional Components document also had some useful enhancements suggested for inclusion. VOMS and PERMIS profiles had been posted to the WG mailing lists and some enhancements were suggested to the VOMS profile. At the end of the meeting, the dates and times for the next teleconferences were set. OGF19 Prof Chadwick attended the following sessions: Prof Chadwick also chaired the OGSA-Authz-WG meeting. SAGA Security meeting On Monday morning, the SAGA group had a joint meeting with security experts to gain feedback on their current API. This was a useful meeting as it allowed the security experts who were present to help to improve the SAGA API definitions, by making the security calls more generic and high level, rather than technology dependent. For example, instead of making a call to add a user to an access control list, this was changed to assign a particular access permission to a user. This would allow more flexibility in the security infrastructure and different technologies to be used as they become available, without affecting the Grid applications (API callers). Page 71 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent A new version of the API was to be produced, incorporating the proposed changes. OGSA-Authz-WG Meeting Prof Chadwick chaired this meeting, during which good progress was made. An editor was found for the remaining deliverable, dates were arranged for teleconferences up to the next OGF meeting (OGF20), and there was a presentation by Takuya Mori of NTT about the first use of the OGF XACMLSAML draft profile that was written by Prof Chadwick. The OGF profile is based on the OASIS XACML-SAML profile. It was stated during the meeting that since the OASIS profile had been written, OASIS had deprecated its use, believing (wrongly) that no-one was using it 2. Takuya and Prof Chadwick therefore agreed to update the OGF profile in the light of the practical experience and OASIS’s decision. The meeting ended with Von Welch standing down as joint chair, being thanked for his outstanding contribution over the years, and a call being made for a replacement. OGSA Security Area Meeting The OGSA Security Area session was designed to bring everyone working on security within OGSA together, to synchronise their charters and goals. As an example, the High Performance Computing group has published its profile and has requested input to its security section, but few people have reviewed this document to date. The meeting wanted to ensure that common texts on security related topics could be produced for use in other OGSA documents. Then when a document was said to be OGSA compliant, it would be clear what it means from a security perspective. Federated Identity All Tuesday was taken up with four sessions on Federated Identity. There were many presentations which reviewed the entire landscape in this area, starting off with an overview by Ken Klingenstein. The sessions covered authorisation and privilege assignment as well as federated identity. The use of Shibboleth was continuing to expand, and several projects were linking Shibboleth to Grids, for example, EGEE (SWITCH) and SHEBANGS (Manchester University). Signet and Grouper were continuing to be developed as a means of assigning privileges to users, but Signet was not a role-based system. Rather, it assigned permission to users to access particular resources, along with conditions of access. Signet was not plugged into any Grid or application or PDP system. This currently had to be done by users of the technology. A set of standard connectors was required if Signet/Grouper were to be used to maximum advantage. These four sessions proved to be a good overview for anyone new to the area, who wanted to gain a wider appreciation of the work that was currently being undertaken. LOA BoF The primary security meeting on Wednesday was the LoA BoF, chaired by Ning Zhang of Manchester. She started the meeting by presenting an overview of LoAs and highlighting the work done by the White House’s Office of Management and Budget (OMB) and NIST. It was noted that there are three separate aspects to LoAs, as follows: 1. The raw data/attributes which describe the authentication process carried out by the IdP 2. The algorithmic process for computing a LoA from this raw data 3. A trust mechanism to assure that the LoA (or raw data) provided by an IdP is correct It was generally agreed that the four levels specified by NIST were good ones. It was also agreed that the whole LoA process was very similar to the certification and assurance process of Grid CAs. Whereas the IGTF validates that every Grid CA is operating to a specific level of trustworthiness, and thus has a binary level of authentication (pass or fail), the NIST LoA specifies four levels. The majority of the meeting agreed that the current IGTF method is insufficient and that both higher and lower levels are needed to cater for existing Grid users. It would, therefore, be possible, when evaluating new CAs, for the IGTF to assign them a NIST LoA at the same time. (This roughly corresponds to point 3 above.) It was pointed out that there was a difference between the trust you could have in an identity, and the trust in authentication of that identity. The meeting noted that there were currently holes in the NIST LoA specification, for example, it does not cater for how the LoA is transported from the IdP to a remote SP. The meeting agreed to request a research group be formed to investigate the holes in the current LoA specifications, how Grids might use the LoA, and what criteria would be used to assess the LoA of an IdP. Ning Zhang agreed to chair this RG. Mike Helm agreed to gather use cases for LoAs. The BoF also agreed to add a work item to the proposed OGSA-Authn-WG to cover the transporting of LoAs in protocol. The definition of the various LoA levels would be done by the IGTF. CAOPS-WG The CA-OPS meeting took place on Thursday. The international group of CAs was continuing to expand, with two South American countries in the process of being added. There was still an issue over how the CA operators can securely communicate with each other, for example, if one of the CAs was compromised. Terena had started to create a PGP circle of trust, but dissenters preferred an S/MIME circle to be created. There were pros and cons for both mechanisms, and this issue was likely to drag on for some time. Rachana gave a talk about validating proxy certificates and attribute assertions, and said that the Globus team was defining an Assertion Validation Service, which, in fact, was no different to the Credential Validation Service which had been specified by Prof Chadwick in an OGSA-Authz-WG draft a year earlier. There was a presentation from the UK NGS about auditing RAs and its value. Some participants were sceptical about the cost-value proposition for this audit. OGSA-Authn Charter BoF 2 A subsequent email to the Authz group from Anne Anderson of Sun, editor of the OASIS XACML-SAML profile, said that this profile had not been deprecated. So there were no restrictions on the OGF WG producing a Grid profile based on it. Page 72 This BoF happened on Thursday afternoon. It was agreed to develop basic use cases in the short term to refine the use of authentication tokens (including un/pw) over secure channels in the OGSA context. 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent Richer use cases would be developed in the long term. There was a long discussion of about whether delegation should be within the scope of the group or not. The decision was to include impersonation and the protocols for transferring tokens within the Authn group and leave delegation of authority to the Authz group. The BOF succeeded in getting editors for four of its proposed deliverables. The more complex or long term deliverables were still awaiting editors. 3.6.7 OGF20 OGSA-Authz-WG This group met on the morning of the first day, and was chaired by Prof Chadwick. The meeting was very productive, as the minutes of the meeting show. The minutes are available at: http://forge.gridforum.org/sf/go/doc14495?nav=1 There were some good presentations by three groups of researchers. However, the WG needed to address the serious issue of how to get people to contribute and progress the WG documents forward between meetings. At that time, there was very little documentation activity between meetings, and it appeared, very few implementations of the existing protocol specifications. Unless this situation was to change significantly by the end of the year, Prof Chadwick suggested that serious consideration be given to closing or suspending the OGSA-Authz-WG. LOA-RG This group met after morning coffee on the first day. This group was tasked with producing use cases and risk analysis for the use of Levels of Authentication Assurance (LoA), and a gap analysis of the current LoA standards. The meeting reviewed a series of use case scenarios and the audience contributed towards a better understanding of them. LoA was an important security concept, and the more understanding and use that implementers made of it, the more secure Grids would become. CAOPS-WG The CAOPS-WG met on Tuesday early evening. One problem the CAOPS-WG was facing was that the commercial CAs were not doing their job properly, and would either authenticate anyone with any name, or not authenticate a person with a name they already possess. Given that the only role a CA had was to bind a name to a public key, this was a major failing of most commercial CAs (caused, incidentally, by their eagerness to shed as much legal liability as possible for their actions). Consequently, the CA-OPS-WG had to devise mechanisms whereby relying parties could control which namespaces they would accept from which CAs, so as to ensure that Grid users were only given their “correct” DNs. Prof Chadwick wondered if there were any value at all in having a certificate from a commercial CA, unless it was badged for your own organisation. Your own local CA could be just as reliable and trustworthy at binding names to public keys. Astronomical Virtual Observatory On Wednesday, there was a full day workshop of Astro Grid. Prof Chadwick met with the organisers prior to the meeting in order to better understand their security requirements, and he then gave a presentation about how existing Grid security software could be tailored to meet the challenges that these raised. In particular, the virtual observatory had a dynamically changing set of member organisations and dynamically changing memberships within those organisations. However, the VO service providers did not want to be continually changing their access control policies, meaning that role-based access controls were needed, with a fixed set of trusted role administrators. It also meant that dynamic delegation of administrative authority to a larger more frequently changing pool of administrators from the VO partner organisations was also needed. The dynamic administrators could then assign VO membership roles to the individuals within their organisations, without any need to change the service providers’ policies. Prof Chadwick highlighted how VOMS and PERMIS could be integrated together to provide one solution that would meet these needs. OGSA Express Authentication Security Profile Wednesday afternoon saw a meeting of the express authentication security profile from the OGSA WG. In Prof Chadwick’s opinion, whilst this work was valuable, it did not go far enough. It would only allow a Grid service provider (SP) to know who you were, without knowing what you were entitled to do. The profile was only half a fix for the real problem. However, it should not be too difficult to extend the current work to provide a fuller more complete solution by borrowing some of the ideas used by Microsoft Cardspace and Trust Negotiation, in which SP’s would publish the authorisation tokens that they required, and IdPs would publish the authorisation tokens they could issue. In this way, users should be able to easily marry the two together in order to ensure that they could get access to the resources they are entitled to. GridNet2 Meeting On Thursday morning, Prof Chadwick took a break from the OGF20 meetings to attend the GridNet2 Advisory Board meeting at the University of Manchester. 3.6.8 OGF21 Prof Chadwick attended a number of working group meetings as described below. Another important benefit of attending this particular OGF meeting was to have F2F discussions with Valerio Venturi of INFN, Italy, who was the main developer of the current VOMS software. Prof Chadwick has two UK JISC-funded projects, VPMan and Shintau, which will require PERMIS to interwork with the VOMS SAML attribute issuing service that is currently being developed by INFN. At the F2F meeting, Prof Chadwick and Venturi discussed in detail how the protocol should be finalised so as to ensure seamless interworking between the two systems. This F2F meeting would lead to new input into the draft OGSA-Authz specification that Valerio has recently written. OGSA-Authz WG Meeting Prof Chadwick chaired this meeting which was well attended. Prior to the meeting, the ADs had discussed with Prof Chadwick the possibility of closing the OGSA-Authz- WG down in the near future, due to the lack of participation by the OGSA-Authz-WG Page 73 3 GridNet2 Reports 3.6 Prof David Chadwick – Kent members. As it turned out, this was one of the most productive WG meetings that the group had during the last 18 months. There was plenty of support from the attendees in keeping the OGSA-Authz-WG going, especially from OMII-Europe. There are several reasons for the lack of support during the last year or two, such as the working group being ahead of user requirements and demand, or researchers not having any funding to work in this area. This was perhaps changing now that Grid usage was increasing, and the scalability limits of existing systems, such as Grid mapfiles, was becoming increasingly apparent. Prof Chadwick presented the latest developments in the two existing authorisation protocol profile specifications (client-PDP and client-CVS), and Valerio presented the first draft of the third and final protocol profile specification (client-CIS (Credential Issuing Service)) that was needed to allow a full set of interactions between the PDP, PEP, CVS and CIS of an authorisation infrastructure. This third profile specified how to use SAML for retrieving the credentials (signed attribute assertions) of a Grid user from a CIS (or Attribute Authority). Valerio also presented the latest developments with their prototype implementation of a CIS, with which researchers at Kent had already been experimenting in order to integrate VOMS with their PERMIS authorisation system. Prof Chadwick presented the latest developments in the VPMan and Shintau projects, and gave an overview of the conceptual model for aggregating attributes from multiple IdPs. The meeting concluded by agreeing that the publication of the three protocol profiles should be treated as a matter of urgency so that they can be reviewed and published for use by the wider community. Security Area Meeting Mike Jones, director of Identity Partnerships at Microsoft, spoke about CardSpace and its likely evolution. He also provided a demonstration of the current system. CardSpace is bundled with MS Vista and a plug-in for MS XP can also be obtained. MS had agreed to cooperate with OpenID to ensure interworking between the two systems in a future release. This might eventually have a significant impact on Grid systems, but in the medium term, it was more likely to affect campus networks and Shibboleth systems. The meeting concluded with a quick overview of the security work that would be covered during the remainder of the OGF21 meeting. CAOPS-WG Session The meeting discussed the Grid Certificate Profile which had just finished its public comment period. A new version of the document was expected to be available by Nov 6, and that version would address all the comments received. The meeting also discussed the Audit document which described a general framework for auditing CAs to ensure that they behave as expected. The group then spent some time discussing the name constraints that Relying Parties might wish to place on certificate path processing procedures. The proposal had wildcards in the specification which was a way of specifying DIT subtrees. Prof Chadwick Page 74 pointed out that the X.509 and X.501 specifications already had ways of defining general subtrees which did not include wildcards, and he forwarded the relevant text to the list. The final topic on the agenda was the use of OCSP servers. It seemed that the requirement for OCSP was dwindling as most RPs seemed to be happy with CRLs or short-lived certificates. Prof Chadwick mentioned the recent WebDAV scheme they had implemented which used Web servers and state based URLs in certificates to provide instant revocation notification and low processing overheads. CAOPS-WG – IGTF Meeting This meeting discussed the current work of the International Grid Trust Federation, and how it was evolving and continually growing. New CA members had been added, Rumania, for example, and other countries, such as Taiwan, were likely to be added very soon. David Kelsey gave a presentation about trust in authorisation, and said that it was now time to start looking at validating the trustworthiness of VOMS servers that issue authorisation credentials, since people were still wrongly thinking that trust in a CA is sufficient to mean that an AA is trustworthy. Prof Chadwick found this to be a refreshing breakthrough in that people were now finally beginning to realise that trust in authorisation was a completely separate layer to trust in authentication, and that it needed to be managed just as rigorously and carefully as trust in CAs. Furthermore, technical measures were needed for enforcing this trust in just the same way as technical measures were needed to enforce the trust in CAs. Fortunately, this was something that Prof Chadwick had been working on for many years, and consequently his group had already built trust enforcing mechanisms into the PERMIS authorisation infrastructure. GridNet2 Workshop At this workshop, each UK participant described the work that they were doing as a result of GridNet2 funding. Prof Chadwick described his work within the OGSA-Authz-WG over the last 4 years. LoA Research Group Mike Jones from Manchester presented the results of the Level of Assurance/Authentication (LoA) survey that they had recently carried out. This showed that most respondents regarded LoA as an important or essential requirement in federated environments, and that tools to support this should be made available to the community. The utility of the LoA had already been effectively demonstrated in projects such as FAME-PERMIS, run by Manchester University. In a new FW7 Integrated Project, that was due to start on 1 January 2008, Prof Chadwick would be developing tools to enable SPs and IdPs to compute the LoAs of authentication sessions, so that they can be used in authorisation decision making. Express Authentication Profile Duane presented the latest developments in specifying an SP’s policies for authentication and message encryption. Prof Chadwick raised the issue that authorisation still was not being covered, and that this was equally important in order for a client to 3 GridNet2 Reports 3.7 Mr Xiaoyu Chen – Brunel be able to establish a successful session. The area director Blair Dillaway agreed, and further stated that the current drafts were worrying in that they strongly coupled the policy specification to the transport mechanism via EPRs. He suggested that these should be de-coupled, and the document authors appeared to agree to this (as does Prof Chadwick), so Prof Chadwick hoped that the next drafts would do this, as well as indicating how authorisation policies could also be added. Overall Comment The level of attendance at OGF meetings was falling. Prof Chadwick considered that perhaps this reflected the lower importance researchers were placing on the production of Grid-specific standards and profiles, or perhaps on the slow pace at which the OGF standards were produced. Either way, it was a cause for some concern, and Prof Chadwick was starting to question whether attending these meetings was a cost-effective use of time and resources. He suggested that other standards forums such as IETF, Liberty Alliance and OASIS might be more productive. 3.6.9 OGF22 OGSA-Authz-WG Meeting Prof Chadwick chaired the working group session, which proved to be very productive. The group agreed an aggressive schedule for producing its final deliverables (protocol profiles) for public comment, and closing down the OGSA-Authz-WG at the Barcelona meeting (OGF 23) or, at the latest, the one after that. The minutes of the meeting have been posted at: https://forge.gridforum.org/sf/go/doc15093 The major contentious issue was what to do about standardising the attributes and obligations that are passed in the current protocol profiles, if the current WG is to close down. After a long discussion, an impasse was reached, so it was likely that nothing would happen in the short term, until enough people came together to push for either a new WG or an update to the current Authz charter. Either way, it was felt that this would be a long term effort as it would be based on the requirements of actual user groups who were actively using the profiles that had already been specified. Demonstration of OGSA-Authz Interworking Several members of the Authz group demonstrated interworking between different products based on the OGSA-Authz specifications. These were GT4, PERMIS, VOMS, GP-BOX, and Sun’s XACML PDP. This shows that the profile standards produced by the OGSA Authz working group are now maturing and are ready for wide-scale adoption. Security Area Group Meeting This meeting highlighted the security work that was being planned during the week, as well as summarising the output of the OGSA Authz meeting that had met earlier that day. It is clear that overall the amount of effort that is available to push security standards/research topics forward at the OGF is less than the number of topics that could be explored. This is a pity, since the OGF has done some good work in helping to ensure interworking between different security component providers, but in some respects it has been technology push as the application developers have either not seen the need for the advanced security tools that have been made available or have made do with simple quick fix and/or home grown solutions. Other Meetings There were various other general interest meetings during the week, such as the Town Hall Meeting to discuss future directions of the OGF, an invited talk on Cloud Computing and the plenary by Charlie Catlett. As always, a valuable part of attending the OGF was the ability to network with other researchers whom it would otherwise be more difficult to meet. 3.7 Mr Xiaoyu Chen – Brunel Mr Xiaoyu Chen is based at the School of Engineering and Design, Brunel University. For a summary of Mr Xiaoyu Chen’s GridNet2 activities, see section 2.7, Mr Xiaoyu Chen – Brunel. 3.7.1 OGF21 Mr Chen was co-chair of the OGF-UR group and a member of the OGF-RUS working group and participated in the sessions of these two working groups at OGF21. UR-WG Session The UR-WG was being re-chartered and aimed to have recommendations for UR version 2.0 at forthcoming OGFs. The content of the UR-WG session was divided into three main parts: • Group re-chartering • Review of UR 1.0 • UR 2.0 perspectives The new chairs of UR-WG were Donal Fellows (University of Manchester) and Mr Chen initially, and there were plans to update the charter soon. The Usage Record 1.0 (UR 1.0) had been widely accepted as a standard usage presentation schema and was being used in many production Grid projects including EGEE/WLCG, UNICORE, SweGrid and etc. With wider adoption and experienced implementations, it had been observed that UR 1.0 lacked some elementary usage properties, such as general VO and executing site property. Besides, the UR 1.0 concentrated on the computational usage metrics of a single batch job without fully or really supporting data and network usage representation, which were extremely desirable within UK e-Science Grid. It was therefore proposed to develop UR 2.0 so that the usage record representation was updated as follows: • To be a flexible information model (aligned with OGSA information model) • To support computational, data and network resource usages Page 75 3 GridNet2 Reports 3.8 Mr Andrew Cooper – OERC • • To extend usage representation for non-trivial jobs (for example, a workflow job or service that might stage a couple of batch jobs) including computational, data and possible network usages To have summary usage representation, which is critical for RUS usages to reduce the number of usage records It was recommended within the session that an experience document would be a reasonable starting point. However, more considerations were needed to format a strategy. RUS-WG Session The RUS-WG focused on discussion of the draft of the RUS Core IDL specification. In the pre-release draft version, RUS operations were refined and generalised for service configuration, usage population, usage query, usage modification and auditing. Two controversial operations, RUS::replaceUsageRecords and RUS::extractUsageRecordIds, had been removed from the RUS Core IDL specification and replaced by general modification and extraction operations. Further work needed to be done quickly for rendering the RUS IDL specification, initially based on WS-I profile and WS-RF profile. The deadline for WS-I profile rendering was the first week of December, 2007. The draft document was submitted to the chairs. The draft document is available at: Pordes (OSG), Stephen Pickles (UK NGS), Stephen Newhouse (OMII) and others. The main contribution of Dr Delaitre’s team was within the job submission area and consisted of addressing interoperability issues on multi- Grids using a portal-based solution. The advantage of a portal-based solution is to hide the different middleware and have a technology-neutral solution which allows e-Scientists concentrate on their research. Most scientists have run their applications on one Grid and there was little support for them to run the same applications in a multi-Grid environment. Dr Delaitre’s solution was offered as a service through the multi-Grid GIN resource testing portal, which aims to support major Grids, including OSG, TeraGrid, NGS, NorduGrid and EGEE. The GIN-portal supported the first four Grids where a workflow could be designed and components of the workflow could be executed on these different Grids with support for different certificates. Some collaboration work started with Stuart Martin from Globus following Dr Delaitre’s visit to Argonne and Fermilab in June 2006. Between June and September 2006, a GIN resource testing portal was deployed and tested with resources on TeraGrid, OSG, UK NGS and EGEE. The main objectives of the GIN resource testing portal, which were disseminated at GGF18, are listed below. • To allow members of the GIN VO to test GIN resources http://forge.ogf.org/sf/go/artf6090?nav=1 • To verify integrity of different infrastructure 3.8 • To check interoperation of different and multiGrid infrastructure Mr Andrew Cooper – OERC • To provide repeated and user interactive testing Andrew Cooper’s report is incorporated into Dr Andrew Martin’s report. • To run applications on a multi-Grid environment For more information, see section 3.26, Dr Andrew Martin – Oxford. These main objectives were presented in more detail at the GIN-CG. For a summary of Mr Andrew Cooper’s GridNet2 activities, see section 2.8, Mr Andrew Cooper – Oxford. Dr Delaitre had a number of discussions after the GIN presentation with key members, including Gabriele Garzoglio from OSG, about the GIN resource testing portal and interoperability issues on different Grids. Dr Delaitre carried out a number of live demonstrations of the GIN resource testing portal at Washington by demonstrating execution of multi-Grid applications on OSG, TeraGrid, UK NGS and EGEE. Dr Delaitre had discussions with Globus about the MDS infrastructure used for testing the GIN resources. Dr Delaitre also talked with Ruth Pordes about experiences with the P-Grade GIN resource testing portal to run a biology application on a multi-Grid environment. Dr Delaitre reviewed the work done so far on the GIN resource testing portal with Stuart Martin and devised future plans about the GIN resource testing portal. 3.9 Dr Thierry Delaitre – Westminster Dr Thierry Delaitre is Systems and Services Manager at the School of Informatics, University of Westminster. For a summary of Dr Thierry Delaitre’s GridNet2 activities, see section 2.9, Dr Thierry Delaitre – Westminster. 3.9.1 GGF18 GIN-CG Dr Delaitre presented work progress on interoperability issues on multi-Grids using a portal-based solution at the GIN-CG. The GIN-CG was set up to explore the interoperation and interoperability issues in running and using global Grids, and contributed to the broader Grid community in the context of the e-Science Function of the OGF within the area of Grid Operations. The main directors from most Grids attended the GIN-CG session, including Dane Skow (TeraGrid), Ruth Page 76 Attending the GGF18 was very beneficial for Dr Delaitre in relation to the GIN-CG, Westminster and NGS. Experiences gained through this multi-Grid interoperability work were very valuable to the community and to Dr Delaitre’s work at Westminster as an NGS partner. 3 GridNet2 Reports 3.10 Dr Abdeslem Djaoui – STFC 3.10 Dr Abdeslem Djaoui – STFC Dr Abdeslem Djaoui’s reports are consolidated into Steve Fisher’s reports. For more information, see section 3.12, Dr Steve Fisher – STFC. For a summary of Dr Djaoui’s GridNet2 activities, see section 2.10, Dr Abdeslem Djaoui – STFC. 3.11 Mr Donal Fellows – Manchester Mr Donal Fellows is software engineer at the Research Computing Services, University of Manchester. For a summary of Mr Fellows’ GridNet2 activities, see section 2.11, Mr Donal Fellows – Manchester. 3.11.1 OGF21 and OGF22 architecture for execution and management of jobs on a higher-level Grid that abstracts the details of its implementation away from the model that it presents to users. The work in OGSA also encompasses work to try to “cross-fertilise” best practice and lessons learned from one group to another, which greatly multiplies the effectiveness of the groups concerned. Outcomes during GridNet2 Mr Fellows made significant contributions to the following documents: • The RSS group (co-chaired by Mr Fellows) moved a document through its public comment stage and is working on resolving the comments. • The JSDL group prepared an errata release to the JSDL v1 specification, and published extensions for HPC integration and Parallel Applications. • The OGSA group published the following documents: Mr Fellows used his funding to attend two OGF meetings (OGF21 and OGF22) and two OGSA F2F meetings (at Fujitsu America in Sunnyvale, CA, in August 2007, and at Imperial College in London in January 2008). This enabled him to participate significantly in the following OGF working groups: • A glossary of Grid terms, • A document on how to model entities on the Grid and draw the models together • An updated description of usage scenarios for specifications in the OGSA space for the execution of non-trivial jobs • Open Grid Software Architecture (OGSA) • Job Submission Description Language (JSDL) Responsibilities • Usage Record (UR) Mr Fellows took on the following responsibilities: • OGSA Resource Usage Service (RUS) • Co-chair of the UR-WG • OGSA Resource Selection Service (RSS) • Editor of the Reference Model • HPC Profile (HPCP) 3.12 • Reference Model (RM) • GLUE • Grid Scheduling Architecture (GSA) • Grid Resource Allocation Agreement Protocol (GRAAP) Mr Fellows also attended the ET-CG to help work towards a common core syllabus for teaching people about the Grid, which was of key importance for making Grid Computing a long-term research field. The major themes of Mr Fellows’ activities within OGF were split into four areas: 1. Description of Grid Entities, which encompasses his work within the JSDL, RM and GLUE groups Dr Steve Fisher – STFC Dr Steve Fisher works in the Particle Physics Department at the STFC’s Rutherford Appleton Laboratory. The following report includes the activities for Dr Abdeslem Djaoui (Rutherford Appleton Laboratory) and Mr Mark Leese (Daresbury Laboratory), who were also funded by Dr Steve Fisher’s GridNet2 award. For summaries of the GridNet2 activities covered by this award, see the following sections: • Section 2.10, Dr Abdeslem Djaoui – STFC • Section 2.12, Dr Steve Fisher – STFC • Section 2.24, Mr Mark Leese – STFC 2. Discovery of Grid Entities, which encompasses his work within the RSS, JSDL, HPCP and GSA groups 3.12.1 3. Accounting of Grid Entities, which encompasses his work within the UR and RUS groups Dr Fisher, Dr Djaoui and Mr Leese all attended GGF15. Dr Fisher and Dr Djaoui also attended the INFOD-WG that was held immediately prior to GGF15. 4. Architecture for Higher-Order Grids, which encompasses his work within the OGSA, GSA and GRAAP groups The work within the OGSA and GSA groups at the architectural level is part of drawing the other strands together to produce a significant part of the GGF15 INFOD-WG – October 05 An INFOD-WG F2F took part on the 2 and 3 October and was hosted by Oracle. The two days were spent discussing the specification and how to respond to Page 77 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC comments on the complexity and size of the specification. It was agreed that the specification should be split in two or three smaller parts with the first one being the base specification. Dr Fisher and Dr Djaoui contributed ideas on how to make this base specification an extension to existing notification specifications, adding a registry type functionality for metadata, but not replicating the publish/subscribe paradigm which is already covered by more than one specification. Progress Dr Djaoui made a presentation in the INFOD session on the new base specification interfaces (see https://forge.gridforum.org/sf/go/doc6337) and Dr Fisher took the minutes. There was a good level of interest from those who do not normally attend INFOD-WG meetings as well as from the regulars. The link between INFOD and WSN was discussed. Following a short update on development of the V2 schemas, Mr Leese presented a list of known open issues. A selection of these issues follows: This was the first report. NM-WG (Leese) The NM-WG held two detailed design discussion sessions at GGF15, which Mr Leese ran as NM-WG co-chair. There was attendance and some input from industry representatives, including Cisco and AT&T. • Schema Normalisation Depending on the scenario, there were differing data items which could be abstracted out of the data section of a schema and into the metadata. Evaluation If a system was returning a lot of data, all of which had the same timestamp, the system should not include the same timestamp for each piece of data. By extension, it would be better to pull out certain pieces of data and refer to it just once, most likely as metadata applied to many pieces of data. Measurements did not need to be grouped by something other than the current metadata. The F2F meeting was very useful. After the GGF session everyone was finally convinced that the specification must be broken down into manageable chunks. This was an important step. Further Action Work on breaking up the specification. Progress This was the first report. The group believed there were four options for further consideration: OGSA-WG (Dr Djaoui) • Do not normalise but just zip/compress all data to reduce the overhead of retransmitting unchanged data • Define just metadata in a schema • Use RDF (Resource Description Framework) • Consult the wider Web Services community Dr Djaoui attended most of the sessions by the groups related to the OGSA-WG, such as BES and Data and Information Model. In particular, Dr Djaoui was asked by the OGSA-RSS-WG to be present for their presentation where discussed the relationship of RSS with information services was discussed. This discussion created a better understanding of where the boundary between the two lay and there was agreement to continue discussions in the future. • NM-WG planned to use an extension to the schema’s namespace. In the current format, this is available at: On the 6 October, Dr Djaoui took part in the OGSA-WG F2F which was hosted by IBM. Dr Djaoui made a presentation on some changes needed in the OGSA document version 1.5. Following a discussion, these changes were accepted and Dr Djaoui was asked to update the document. Dr Djaoui also explained how INFOD (with the registry functionality) could be leveraged for the OGSA architecture, and the importance of giving the OGSA requirements to INFOD before the specification is finalised. During the Information Model discussion, when the GLUE schema versus CIM was again being discussed, Dr Djaoui argued that the fact that GLUE was widely used in deployed Grid should not be ignored and that GLUE and CIM were not irreconcilable. Dr Djaoui suggested that the Information Modelling group should collaborate with the GLUE team, so they could work out a mapping from one to the other. Evaluation There was some progress on bridging the gap between INFOD and OGSA Further Action The GLUE/CIM relationship should be followed up. Page 78 Schema versioning http://ggf.org/ns/nmwg/base/VERSION A GGF document on namespaces (Standardised Namespaces for XML infosets in GGF) was about to be published and was to be consulted. Versions were to be specified using major and minor numbers. Versioning should be explicit, so that a receiving system could decide whether it supported the functionality required by that version of a schema. No default version should be implied when version information was missing or incomplete. • GridFTP logs It should be possible to express GridFTP logs using the schemas. This would be of interest to users or network researchers requiring an application rather than raw network view of network performance. An NM-WG schema representation would assist development of a data gathering mechanism. 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC • Validation Authority Approval/compliance was indicated by placing a schema in the NM-WG namespace. When a new tool was developed, or a group proposed a new version of a tool or characteristic specific schema, how would the new/modified schemas gain NM-WG approval? Mr Leese was concerned that the NM-WG should define an approval mechanism very soon. The third NM-WG session was dedicated to updates on projects making use of the V1 and draft V2 schemas, including demonstrations. Mr Leese presented current European efforts, including the performance monitoring component of EGEE-JRA4 (Development of Network Services) based at NeSC and EPCC, plus his own gridmon work for GridPP and the UK NGS. It was particularly pleasing to demonstrate the EGEE-JRA4 Mediator and associated Diagnostic Tool. The Mediator had been a significant proof of concept for the NM-WG, since it could present a single (NM-WG) interface to data consumers to several underlying NM-WG-compliant network monitoring infrastructures. The Mediator was again demonstrated obtaining data from different networks and monitoring infrastructure, including the GEANT (DANTE), Abilene (Internet2) and ESNet networks via the perfSONAR infrastructure, this time with the added benefit of data graphing using the Diagnostic Tool. The group held its outreach session on Tuesday 4th October. Jason Zurawski presented an update on the schema work, followed by Eric Boyd and Mr Leese providing abridged versions of their perfSONAR and European update talks from the previous day’s “demonstrations” session. While the NM-WG had varied WS exposure, active group participants were, in the main, networking experts, not software engineers. The group felt that it needed to fast-track the development of its WS expertise, so that the group could develop Web Services solutions making use of the NM-WG schemas without distracting from the group's core objectives: applying the group’s networking knowledge to the schemas themselves. To this end, Mr Leese recruited Marlon Pierce and Shrideep Pallickara from the Community Grids Lab at Indiana University to run a half-day overview of Web Services, culminating in a lengthy Q&A session. The session was run as a community activity since it appeared that the NM-WG was not the only group in this situation. Evaluation The NM-WG has made progress through productive F2F meetings. The value of meeting F2F and intensively working through a pre-agreed set of issues should not be underestimated. Experience had shown the group that this was far more productive than telephone conferences and e-mail discussions. Holding F2F meetings appeared to focus everyone’s mind on providing the information, opinions and arguments that were required to make the necessary decisions. The US group members were updated on European work and vice-versa, keeping the work in-step and facilitating knowledge sharing. The wider Grid community were updated on what the NM-WG was doing. Further Action The group planned to produce a set of common tool and characteristic specific schemas (for example, ping, iperf and traceroute) which would satisfy the majority of users needs, at least in the short-medium term. Progress This was the first report. NM-WG Community Activity (Leese) Web Services Performance: Issues and Research The NM-WG was taking advantage of the near ubiquitous nature of XML and Web Services (WS) to exchange data. XML provided a convenient syntax for building the data structures required for the information the group wished to exchange, while WS provide platform and programming language independence. WS were also a sensible interface to use for inter-working with Grid systems, such as resource brokering middleware, which may, for example, use network performance information in selecting the Compute and Storage Elements used for running Grid jobs. In general, the NM-WG did not find data transit to be a problem, but processing XML on the end machines, namely the relatively large size of the NM-WG XML documents compared to the amount of useful data they encapsulated and the memory requirements of large DOM-based trees, was a cause for concern. The group was aware of performance improvement techniques, such as binary encoding of XML, and made steps to alleviate problems, such as separating data and metadata so that the metadata need only be transmitted once, and can then be referred to using a single metadata ID field. While this session asked the right questions, the answers were far less forthcoming, and there were differences of opinion between the commercial and academic speakers. The workshop did reveal that XML (de)serialisation was the main performance bottleneck, since XML markup was verbose and, as a result, computationally expensive to generate and parse. NM-WG would continue with Web Services, as the benefits still outweighed the disadvantages. It was hoped that the performance problems would diminish through a mix of WS technology reaching maturity, hardware speed increases and the implementation of intelligent caching of recent data. Evaluation NM-WG learned that even the Web Services experts did not have all the answers to WS performance, and that the group needed to press ahead. Further Action The group planned to carry on with WS. Progress This was the first report. Page 79 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC 3.12.2 needed to extend interest with Grid researchers and developers. GGF16 Dr Fisher attended GGF16 and the INFOD-WG that also took place at the same time. • There was a lot of interest in using plots of network performance to analyse performance problems, for network and Grid operators, and Grid users. Examples showed how useful this could be, but it was noted that some form of guide would be needed for helping to identify or rule out common problems using graphs of performance. There was a lot of interest in such a guide. • Not for the first time, there was some scepticism about network operators wanting to share data. They had traditionally been very protective of their performance data, since it could be commercially and security sensitive. In response, NM-WG had some research and educational operators (ESNet, Internet2 and DANTE) who were ready to share data and actively want to share it. While access to data was an important issue, NM-WG’s work was predominantly concerned with providing a mechanism for sharing the data, not persuading operators to share the data itself. The group’s best option here was to lead by example, complete development of the schemas and put them into service with enthusiastic adopters, so that the benefits could be readily seen in real life scenarios. INFOD-WG – February 06 The INFOD-WG F2F was hosted by Oracle but situated in the GGF hotel. Much of the time was spent going through the use cases to ensure that the draft specification would match them. In the GGF session, Dr Fisher presented the status of the group and its work and plans. Evaluation The F2F meeting was once again very useful. Unfortunately, not being in the US, the GGF was not as well attended as it sometimes was. Further Action The INFOD-WG planned to work on finalising the specification so that the document could be submitted and the 60 day public comment period ended by GGF18. The group also intended to plan implementations, and consider an advanced specification and matching use cases. Progress The group had finalised definition of what goes into the base specification, and decided to adopt WSN Notify rather than inventing something else for the basic notify call. 3.12.3 GGF17 Mr Leese attended GGF17. NM-WG The full NM-WG report is available at: http://nmwg.internet2.edu/meetings/ggf17/nmwgggf17Summary.txt The second session was advertised as a discussion on the following question: The NM-WG schemas are gaining popularity in the research and educational network domain within Europe and the US, for example with EGEE, DANTE and Internet2. What are the barriers to their adoption (or trials) in Asia? During the second session Mr Leese spoke to a Japanese Grid academic, which quickly became a useful and very revealing exercise: • He felt that the problem in Japan (at least) was that there was a large separation between computer scientists and the networking groups, with no available to bridge the gap. He also observed that local networking people did not feel a need to update their operational practices, that is, they were happy to drive network equipment using just a command-line interface, because it was sufficient for their needs. A concept like Web Services would therefore be viewed as “too new”, with unproven benefits, and would garner little interest as a result. • He also felt that while plots of performance were useful, the case for adaptive Grid applications and more inclusive/autonomous network operation was not (yet) sufficiently strong to justify the time required to be more involved, a view he believed would be shared by his networking colleagues. It was hoped that the deployment of a major NM-WG compliant network monitoring infrastructure such perfSONAR would change this perception. • The contact's group would be happy to look at software using the NM-WG schemas, but as always, time and people are in short supply, so it needed to be a “low hassle” exercise. This was defined as software that could be easily installed on a PC. This was something that NM-WG could potentially investigate. NM-WG held two sessions at GGF17, which Mr Leese ran as NM-WG co-chair. In the first session, Mr Leese gave a 90 minute presentation covering NM-WG basics: group history; the group’s previous work (version 1 schemas, including lessons learned); and the current V2 work. The slides for the presentation are available at: http://nmwg.internet2.edu/meetings/ggf17/nmwgggf17AllSlides.ppt The V1 and V2 material was reinforced with overviews of the relevant EGEE-JRA4 and perfSONAR work, complete with simple demonstrations of each. Background was also given as to why this was important work from the point-ofview of both network operators and Grid users. • There were 12 people present for the presentation session. As Mr Leese was the only “core” group member present, this represented better attendance at the general outreach session by non-core members than previous GGFs. This was especially pleasing given the somewhat disappointing turnout at this event. However, several of the 12 were networking or telco professionals, again indicating that the group Page 80 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC • • The contact was not sure how monitoring infrastructures such as perfSONAR interrogated the underlying network infrastructure, but expressed the view that both Cisco and Juniper equipment is popular in Asia, so any software should work with those network boxes. For clarification, Mr Leese pointed out that there was much more to this work than simple SNMP polling of devices. On a positive note, the contact believed the situation was improving, giving the example of a South Korean team who had used Web Services to link an in-house resource scheduler to software produced by another group(s). The Korean team was very surprised at how easy it had been to use Web Services to interoperate disparate software, and were now more receptive to WS options. This all suggested that a more inclusive approach was required to achieve an example deployment in Asia. NM-WG would need to capture a group's requirements and actively assist the group in building software which implements the schemas. Alternatively, a group might be open to deploying an instance/node of an existing infrastructure, such as perfSONAR, if it met the local requirements and a certain amount of support was available. This was an issue for discussion during a forthcoming NM-WG meeting/phone call. As a slight aside, some separate attendees thought that Asia might only take the NM-WG work when it was complete, analyse it, and then adapt it to their own requirements. Other Information Mr Leese also visited the Grid World Japan exhibition floor and spoke with several exhibitors, ranging from services companies such as Fujitsu to OEMs, including HP. There was clearly more commercial adoption of Grid computing in Asia-Pac than US-Europe, yet while the people Mr Leese spoke to knew that the network was essential to bandwidthdependent and large-scale Grid deployments, the impression Mr Leese received was that no one was doing anything novel with the commodity IP networks they were using, or the Grid middleware and applications themselves. The exception was NaReGi, a Japanese academic Grid project, the Grid Network Group of which were working to produce a secure and high performance Grid network middleware. This included protocol research and investigating management of network resources based on Virtual Organisations. NM-WG relevant work would include the real-time measurement of Grid application network traffic (flow-based measurement of RTT, throughput etc.). Mr Leese identified this as another potential door into Asia for NM-WG. Evaluation The NM-WG sessions and GridWorld exhibition area were very useful. The group now had a handle on why there was little interest in their work from Asia-Pac. As a result, the group was in a better position to address the disinterest, and until the group could do so, the group valued the contact with people who provided valuable feedback on the theoretical aspects of their work, but did not have the overhead of producing implementations themselves. Further Action Further action needed to be discussed within NMWG. NM-WG was keen to build presence in Asia-Pac, but was already under-volunteered. Based on the comments NM-WG has received, NM-WG could postpone this until the base schemas were complete (thus freeing up effort) and a stable “real-life” demonstrator was in place (for example, perfSONAR deployed by Abilene and DANTE). At the very least, NM-WG needed to maintain relationships with the new contacts. Progress Since his last report, Mr Leese was involved in developing the NM-WG schemas through group conference calls, which Mr Leese chairs. For the version 2 schemas, the group’s roadmap specified production of a GGF Experimental Document, summarising details of trial V2 implementations and operational experience. This, NM-WG hoped, would naturally lead into a GGF Recommendations Documents, providing a technical specification of the schemas and their intended use. As development of the schemas moved towards a conclusion, with the schemas already having seen some trial deployment within the ESNet, Abilene and GÉANT2 networks, the group felt comfortable moving straight to a full recommendation. NM-WG was planning how best to produce the full recommendation, which was complicated by a need to package and make available the base schemas and pre-defined NM-WG schemas, that is, the base components from which particular performance characteristic and performance measuring tool specific schemas were produced, together with the sample schemas already defined for the most common characteristics and tools. Mr Leese would be involved in producing the recommendation document. Grid High Performance Networking Research Group The work of GHPN focuses on the relationship between network research and the development of Grid applications and infrastructure. GHPN’s work attacks the problem space from both sides, identifying the following issues: • Requirements of Grid applications which were not being met or understood by the networking community • Advanced networking solutions that were not being used by the Grid GHPN had two sessions scheduled for this event. It was unfortunate that the first session was cancelled as late as 5th May. The second session consisted of a series of presentations following the topic of Grid Network Interface. Two of the three talks were particularly interesting in that they described projects aiming to closely interface Grid applications with the network. This was of interest to both NM-WG and ESLEA, the project driving exploitation of the UKLight infrastructure, which includes development of network control plane software for automated provisioning of circuit-switched lightpaths. Page 81 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC • Overview of Grid networking and applications research in Japan, Akira Hirano (Senior Research Engineer, NTT Network Innovation Labs, Japan). Akira spoke about application-driven network capacity-on-demand. This included G-lambda (http://www.g-lambda.net), a project aiming to create a standard Web Services interface between Grid and network resource management systems, with which Grid applications could make ondemand requests for guaranteed bandwidth channels. • Enlightened Computing: Applications-Driven Adaptive Compute, Instrument, and Network Resources Integration, Yufeng Xin (Senior Scientist, MCNC, RTP NC, USA). EnLIGHTened Computing aimed to achieve dynamic and adaptive on-demand use of end-toend networking resources. In a broadly similar way to G-lambda, the project aimed to create an optical network test bed supporting virtual subnetworks which provided various QoSguaranteed end-to-end links, over which Grid applications had some level of control. This would involve development of a novel optical control plane, and associated APIs at the Grid application, network resource management and optical control plane layers. The work was relevant to ESLEA, while the project’s aim of near-real-time feedback of resource (including network) performance measurements to Grid applications and middleware was of interest to NM-WG. The performance issues of Web Services meant that it was infeasible for the NM-WG schemas to be used in the network control plane. Indeed, this was not their purpose; they were designed for sharing data at higher layers, including the network-application interface. Having attended the BoF at a previous GGF, which led to the formation of this group, Mr Leese was pleased to see that there was a good draft of their first document. In particular, it covered issues likely to affect UK e-Science, including middle-box interaction with Access Grid, Globus, GridFTP and high bandwidth, long distance networks, as used by projects such as Reality Grid for linking geographically distributed Grid resources. It was not clear from the meeting however, that the document adequately covered all relevant scenarios. For example, there was no coverage of IP multicast, which can severely impact Access Grid operation. CCLRC was well placed to review the work, since it already hosted a number of e-Science relevant scenarios, including Access Grid, compute clusters in a DMZ and large-scale HPC facilities (HPCx) behind a firewall. Evaluation FI-RG could have an impact on UK e-Science but would be of little interest to non-networkers until the third document (evaluation of middle-box approaches and solutions) was available. Further Action It was necessary to ensure that the first FI-RG document considered all use cases which CCLRC were aware of, providing fresh input if not, with the view that the document evaluating middle-box approaches and solutions could made use of at a later date. The FI-RG was actively seeking more active participants. 3.12.4 INFOD-WG F2F – May 06 Dr Fisher attended the INFOD-WG in May 06. No standards work took place, but the meeting was useful in that it brought the G-lambda and EnLIGHTened Computing projects to Mr Leese’s attention. Mr Leese also notified ESLEA of their existence, in case there was expertise which could be shared. This F2F meeting was hosted by IBM in New York. Once again, the INFOD-WG looked at the specification with regard to the use cases. The INFOD-WG decided that in order to get the specification finalised according to the plan at GGF16, the group would have to concentrate on that and not tidy up the use case document. INFOD-WG would still make the use case document available but not submit it formally. This was by analogy with the WSN practice. Further Action Evaluation NM-WG needed to consider making closer contact with G-lambda and EnLIGHTened Computing to look, at the very least, in more detail at their protocols for network-application communication. Considerable progress was made. The remaining work on the specification was mostly editing. Firewall Issues-Research Group The group planned to tidy up the specification so that the document could be submitted and its 60 day public comment period ended by GGF18. This was mostly just editing and proof reading. Evaluation FI-RG planned to create three documents that covered the following topics: Further Action 1. Recording and classifying common issues which Grid jobs faced when their network connections encounter network middle-boxes (firewalls, application level or VPN gateways etc.) Progress 2. Evaluating available IETF middle-box protocols and functions 3.12.5 3. Evaluating approaches and solutions for addressing the issues raised in the first document Page 82 Almost everything had been decided for the specification. GGF18 Dr Fisher attended GGF18 and the INFOD-WG meeting held immediately before GGF18. At the F2F meeting, the INFOD-WG discussed the following topics: 3 GridNet2 Reports 3.12 Dr Steve Fisher – STFC • The feedback received from the reviewers of the INFOD base specification. This included extensive input from the WSN community. • Progress with INFOD implementations • Preparation for the main GGF meeting • Various pending issues – including the precise role of the disseminator in the INFOD model and the role of “Beacons” and “PO Boxes”. The aim of this was to try to pin down what would go into a further specification. Dr Fisher replaced Susan Malaika as co-chair. Dieter Gawlick remained as the other co-chair. As usual, the F2F meeting was much more useful than the formal GGF session. 3.12.6 INFOD-WG F2F – February 07 Dr Fisher attended the INFOD-WG F2F meeting. This was a kick-off meeting for the INFOD implementation project being led by Knoxville and Oak Ridge. Much of the meeting was spent planning the implementation and sharing ideas. The main value of the meeting was to allow the new people doing the implementation to meet those who had been working on the specification. 3.12.7 OGF19 Dr Fisher attended OGF19. INFOD-WG Due to various travel restrictions, there were no INFOD sessions at this OGF but Dr Fisher gave a status presentation to the data area. SAGA Dr Fisher presented work being done at CCLRC by A Paventhan and himself on a service discovery API. The SAGA group liked the idea and agreed that it would fill gaps in the SAGA specification as an addon package. OGF20 was set as a target to produce a written specification. 3.12.8 OGF20 Dr Fisher, Dr Djaoui and Mr Leese all attended OGF20. INFOD-WG Due to various travel restrictions, there were, once again, no INFOD sessions at this OGF but Dr Fisher gave a status presentation to the data area. He said that there would definitely be INFOD sessions at OGF21 and that progress had been very good, the group having very recently submitted a revised specification. It had taken a lot of work and long discussions to get it finalised. OGSA-WG During this meeting, Dr Djaoui had further discussions with OGSA-WG members on the best way to address composition and coordination of services and the role of asynchronous messaging. OGSA-WG was already looking into workflow in relation to JSDL and collecting use cases. Dr Djaoui opted to attend all the sessions on workflow during the meeting and discussing the status of workflow in OGSA at the F2F on the last day. Dr Djaoui also attended and contributed to the discussions for WS-Naming, information services and the OGSA glossary. On the last day, during the OGSA F2F, Dr Djaoui gave a high level talk explaining what OGSA should do. For details, see: http://forge.gridforum.org/sf/go/doc14513?nav=1 The main features of WS-BPEL and Microsoft WF were discussed and areas for possible work presented. It was agreed that in order to better understand what OGSA should do, OGF should look into "Community Practice Documents" on the use of workflow. This should allow the identification of interesting use cases for OGSA to work on. SAGA Dr Fisher presented a draft proposal for the Service Discovery specification. Some suggestions were made on how to improve it and Dr Fisher was encouraged to produce a revised draft. A. Paventhan (not currently GridNet2-funded) gave a demo of his implementation. 3.12.9 OGF21 Dr Fisher and Dr Djaoui attended OGF21. INFOD-WG Instead of having the usual F2F meeting adjacent to the OGF, the INFOD-WG requested 4 sessions during the OGF. This turned out to be much less effective for getting work done. Nevertheless, the group were able to agree on most parts of the revision to the base specification and get a much clearer idea of what would be in the more advanced INFOD specification. OGSA-WG Although it was only possible for Dr Djaoui to attend 2.5 days of OGF21, he was able to participate in discussions at the "Web 2.0/Grid workflow/Parallel computing" workshop. He had some useful exchanges with Microsoft representatives on the new Concurrency and Coordination Runtime (CCR) and Decentralised Software Services (DSS) technologies and their relevance to Grid computing. The relevance to asynchronous messaging, something Dr Djaoui had been advocating within OGSA-WG and OGF since OGF19, turned out to be particularly timely, as CCR is mainly about making asynchronous messaging and coordination easier to program and, it was said, was allowing a whole range of applications to be developed in a variety of disciplines. The main non-commercial implementation project, with much of the effort from the University of Texas, Knoxville, was reported as moving forward well. Page 83 3 GridNet2 Reports 3.13 Mr Andrew Harrison – Cardiff Afterwards Dr Djaoui talked with some OGSA-WG members about CCR/DSS technologies and how they relate to OGF and OGSA. CCR was a runtime library that was mainly used for multi-core programming (MPI-like), but when combined with DSS, it offered a powerful model for integrating distributed applications. At the heart of CCR is the concept of a port (a kind of message queue), but it is not clear at this moment how this relates to messaging, SAGA or the workflow activities in OGF. It is expected that the proposed Community Practice Documents on service composition would provide the necessary use cases to help define what is needed in OGF. The group agreed to continue to explore the topics of asynchronous messaging and coordination further within the Community Practice Documents activity. Dr Djaoui also attended and contributed sessions on Grids in the IT data centre workshop, Data/Compute Affinity - Focus on Data Caching, and Grid Reliability and Robustness. SAGA Dr Fisher presented a further draft proposal of the Service Discovery specification. This time there were only minor changes suggested and he was invited to prepare one more draft to send around, at which point the group would submit it to the OGF editor. 3.13 Mr Andrew Harrison – Cardiff Mr Andrew Harrison is a research fellow/associate at the School of Computer Science, University of Cardiff. For a summary of Mr Harrison’s GridNet2 activities, see section 2.13, Mr Andrew Harrison - Cardiff. 3.13.1 OGF20 Mr Harrison participated in the following workshops at OGF20: • Secretary and Speaker: Sharing Workflows Workflow Management Research Group (WFM-RG) session • Participant: OGSA Workflow • Participant: e-Arts and e-Humanities: e-Science technologies and methodologies in Arts and Humanities research, sessions 1 and 2 • Participant: OMII Software Overview Sharing Workflows Session Mr Harrison was secretary and speaker at the WFM-RG workshop. This workshop addressed how workflows can be shared between different applications, not so much from the perspective of the standardisation of representation, but in terms of crossing different software environments, sharing “results” and provenance data and building communities through workflow accessibility. Mr Harrison’s presentation introduced the WHIP project, funded by OMII, which recently started in Cardiff. WHIP aimed to provide support for sharing workflows, and more generally artefacts, or digital “things” of any description, across different software environments (for example, Web-based portals and thick clients) and different distributed messaging stacks (for example, WS/SOAP, raw HTTP, Web 2.0 Page 84 technologies such as feeds and blogs, and P2P networks). The range of talks was interesting, throwing up many challenges that are often domain or workflow specific. The outcome, to be delivered at OGF21, was a research document on application scenarios. The aim of this document was to ascertain the possible areas of overlap between domains, and existing workflow engines. Mr Harrison would be working on this document. OGSA Workflow This workshop focused on gathering use cases with an end goal of standardizing workflow description within OGSA. Mr Harrison had followed discussions of this group on the mailing lists, and was interested to see how this workshop went. Most of the discussion centred on choosing a standard for workflow representation. In particular, BPEL4WS was discussed as an appropriate representation, as was the Workflow Management Coalition (WFMC) - a group of companies addressing workflow standardisation through XPDL and UML. XPDL was an abstract version of BPEL and led to BPEL 1.0. Both the use case gathering and the representation discussions seem at an early stage. Mr Harrison is ambivalent about the adoption of BPEL based descriptions. One of the factors for the development of scufl — the language used by Taverna — was the perceived inapplicability of BPEL to scientific workflow. More generally, Mr Harrison believes existing systems are unlikely to re-factor themselves to fit in with an imposed representation. If current systems were replaced by compliant systems, many years of experience, expertise and development would potentially be lost to the community. Furthermore, as was clear from the WFM-RG workshop, workflow requirements are often domain specific and hence enactment-engine representations can reflect this more easily than trying to adapt to an umbrella representation. e-Arts and e-Humanities This was a very interesting BoF meeting in which a number of projects presented their work in integrating e-Science technologies with the arts and humanities. Mr Harrison had a particular interest in this area, having been an artist and theatre designer before becoming a computer scientist. Presentations included recent work at TextGrid in Germany, the Arts and Humanities e-Science Initiative in the UK, work emerging from the Center for Computation and Technology (CCT) at Louisiana State University (including a collaborative project with Cardiff on music information retrieval) and the Associated Motion Capture User Categories (AMUC) project, based in Newcastle. This latter project, presented by Sally-Jane Norman, was particularly interesting. It was investigating cross domain mechanisms and requirements for recording human motion, placing dancers, artists and biomechanics into the same room. A particular point made by Ms Norman (paraphrased) was whether it was actually fruitful to build a meta-language to enable cross referencing and storage of diverse motion capture languages, because in so doing, the peculiarities, which are the strengths of the different languages, were lost in the process. Mr Harrison saw an analogy with the discussions surrounding workflow standardisation. 3 GridNet2 Reports 3.13 Mr Andrew Harrison – Cardiff OMII Software Overview Because of Mr Harrison’s involvement in the WHIP project, he took this opportunity to catch up with the latest developments in the software coming from OMII. The WHIP project aimed for integration with a number of OMII projects. In this context, Mr Harrison attended an initial meeting including four OMII-commissioned software projects — Mr Harrison was representing WHIP — dealing with portal-based communications. The meeting was chaired by Neil Chue Hong. WHIP’s integration with OMII would also enable discussions between Taverna and the Cardiff-based Triana workflow system. Mr Harrison sees closer ties between the developers of these projects as the primary route towards addressing the issues of workflow standardisation. This is a bottom-up, application-centric approach, as opposed to a top-down approach. Systems such as Triana, Taverna and Kepler each had strengths and weaknesses stemming from idiosyncrasies that had developed out of their history and the problems they had been addressing. Building more flexible, applicable and usable workflow systems needed to take these idiosyncrasies into account. In Mr Harrison’s opinion, developing closer links between the developers of these systems was the best way of achieving this. Summary In summary, OGF 20 was highly fruitful for Mr Harrison. It helped him understand what workflow standardisation means and how it could best be achieved. Mr Harrison believed the issue required a bottom-up approach, working with existing systems and their idiosyncrasies. Furthermore, users should have a choice of tools at their disposal. The choice might include a programming language, a workflow representation and enactment engine, as some may be suitable for certain situations but not for others. As a programmer, Mr Harrison could choose a language most suited to the problem at hand. This flexibility should also be available to users who were building and running workflows. Integration and advances in current systems were most likely to take hold through sharing experiences and code. Mr Harrison hoped that the WFM-RG research document on application scenarios, to be delivered at OGF 21, would throw light on particular steps that could be taken to this end. 3.13.2 OGF21 This report provides an overview of Mr Harrison’s activities in research and working groups at OGF21 meeting. It focuses primarily on the WFM-RG session and the broader work that is it triggering, as this formed the basis of his GridNet2 application. Mr Harrison participated in the following workshops: • Secretary and Presenter: Workflow sharing and Interoperability – WFM-RG session • Participant: All the Web 2.0 Workshop sessions. • Participant: OGSA Workflow • Participant: GridNet2 workshop Workflow Sharing and Interoperability Session Mr Harrison was secretary of the WFM-RG, and gave a presentation in the session. The focus of the session was sharing and interoperability. Speakers included Ian Taylor (introduction), Matthew Shields (recap of OGF20 WFM-RG session), Ewa Deelman (overview and results of the NSF Workflow Interoperability workshop), David de Roure (myExperiment), and t others. Mr Harrison also gave a talk which focused on embedding workflows into different environments, and in particular how this can be achieved with minimum disruption to existing systems. Mr Harrison discussed RESTful approaches to this issue and the benefits of such an approach. The session was a success with around 40 participants and provided a useful forum for exchanging ideas, and getting feedback from the OGF community. The last two OGF meetings, for which Mr Harrison has been funded by GridNet2, have shown a marked increase in the activity of WFM-RG and this was the objective of his funding application. Activity in the WFM-RG was complemented by related initiatives in the field, including the NSF Workflow Interoperability workshop, and the Workflows for e-Science proposal being submitted to the e-Science networking call. Particular activities that Mr Harrison was involved in include collaborations based on the WHIP project and approaches to interoperability through embedding workflows, particularly work with the Kepler workflow system and the P-Grade portal. A starting point of these collaborations is arriving at use cases for interoperability. The WFM-RG was continuing work on these use cases with a view to presenting them to the OGF community at OGF22. These use cases would form the basis for developing standardised interfaces between workflow systems. Other Sessions The Web 2.0 workshop sessions were particularly interesting for Mr Harrison because part of the WHIP project was investigating Web 2.0 architectures and technologies for sharing and presenting workflows through Web portals. Notable speakers included Geoffrey Fox (Web 2.0 Grids and Cyberinfrastructure), Savas Parastatidis (Web 2.0 infrastructure and applications) and Marlon Pierce (Real-time Web 2.0: Evolution of Middleware for Grid-based Instruments and Sensors). Web 2.0 represents an interesting growth area for Grid research and is proving to be a challenging and sometimes disruptive influence on the community. Mr Harrison believes this will enhance the debate and quality of research emanating from the community. The OGSA workflow meeting was interesting. The OGSA-WG was taking a flexible approach to integrating workflow into the OGSA architecture, beginning with a survey of capabilities of different workflow systems. Mr Harrison planned to provide details of the Triana workflow engine to the group. The GridNet2 workshop allowed various delegates to share their work in relation to GridNet2 and the Grid community as a whole. It was hoped that this would lead to a more cohesive view of the contributions made by the institutions involved and the UK e-Science community in general. Page 85 3 GridNet2 Reports 3.13 Mr Andrew Harrison – Cardiff Summary OGF21 was very fruitful. The WFM-RG was gaining momentum which was spilling into other projects and proposals. Web 2.0 paradigms were affecting the directions of research in the Grid community as a whole, as well as Cardiff's contributions to the work in the WFM-RG. Mr Harrison requested funding for attendance at two OGF meetings - Manchester and Seattle. His GridNet2 funding covered about 80% of his expenses to Seattle, and he drew on other funds to cover the rest. While the progress over the last two OGF meetings with regard to the WFM-RG was significant, the results were still preliminary and required further efforts to bring them to fruition. 3.13.3 OGF22 Requested Clarifications from Application In applying for GridNet2 funding for OGF22, Mr Harrison was asked to clarify the following questions: • The project that funded Mr Harrison’s current position had enough travel budget for one person to a European destination. • • • This was disseminated to a variety of groups including the WFM-RG and OGSA-WG. Some responses were received promptly, and there were still more responses arriving. Workflow embedding - allowing workflows to run within their own environment, but invoked from another • Development of a meta language - allowing different proprietary languages to be mapped to a single standard one • Semantic annotation/description/classification particularly important for workflow sharing The session was then opened for discussion, which was very fruitful. Six participants offered their services in helping to produce a document that would build on the preliminary analysis of the questionnaire. The aim of this document was to provide a means of understanding the requirements and challenges of the three approaches to interoperability. The different approaches could lead to different solutions, such as APIs and workflow language specifications. It was hoped that addressing the challenge on different levels would result in greater uptake and involvement from the community than a one-sizefits-all approach. Work on the document was begun. Page 86 If integration with the OGSA workflow group was an aim, as was mentioned, how would you go about this? Things had progressed somewhat. The OGSAWG workflow discussions were waiting on the WFM-RG to produce some results in terms of interoperability and sharing standards with the aim of integrating these results. A member of OGSA-WG had kindly agreed to contribute to the WFM document. The WFM-RG session was very well attended, with 34 registered participants. Ian Taylor presented an overview of what the group had discussed in the previous two OGF meetings. • What precisely was Mr Harrison’s intention to achieve specifically at OGF22? Mr Harrison and the WFM-RG specifically wanted to move beyond the discussion phase on workflow interoperability and sharing. The group achieved this through gaining a commitment from a number of members of the community with wide-ranging specialisms and experiences to contribute to a research document on the issues of interoperability and sharing. Workflow Sharing and Interoperability Session Then Mr Harrison presented a summary of the results from the questionnaire. The feedback from the questionnaire showed that there were a number of projects that required interoperability and sharing. The possible approaches to achieving these fell into three main categories: Were there any other sources of funding that might enable participation beyond OGF22? Mr Harrison aimed to use the travel funds available from the current project to fund his travel to OGF23. This report provides an overview of developments in the WFM-RG at OGF22. The week before OGF22, Mr Harrison created an online questionnaire in order to elicit opinions, experiences and requirements from users, designers and developers of workflow systems regarding interoperability and sharing. See: http://bender.astro.cf.ac.uk/wfmrg Were there any other sources of funding? • What are the results that are preliminary? The preliminary results are the dissemination of WFM-RG strategy regarding interoperability and sharing. The WFM-RG spent OGF20 and 21 discussing the possible approaches to interoperability with members of the community. Mr Harrison suggested that there was some consensus, at least on the fact that there were different mechanisms by which styles of interoperability might be achieved. This was a step forward from the view that only a wholesale adoption of some specification (for example, BPEL) was the way forward. • It would be helpful to have more detail about Mr Harrison’s goals and the time frames required to achieve them. The WFM-RG achieved their goal for OGF22, which was to get commitments from members of the community to contribute to a document setting out the possible routes to interoperability. The group aimed to have this document completed by OGF23 where they could present the results. Building on top of this during the following year, the WFM-RG hoped to achieve some tangible guidelines for the community, as well as examples of interoperability and sharing between projects. 3 GridNet2 Reports 3.14 Mr Chris Higgins – Edinburgh 3.14 Mr Chris Higgins – Edinburgh Mr Chris Higgins is the Workgroup Leader, Products and Services Development for EDINA at the University of Edinburgh. For a summary of Mr Higgins GridNet2 activities, see section 2.14, Mr Chris Higgins – Edinburgh. Mr Higgins was funded under GridNet2 from Dec 2005 to engage in activity which would develop synergies between the e-Science/Grid and Geographic Information (GI) communities. This report complements the presentation that Mr Higgins gave at the GridNet2 workshop at OGF21. Copies of this presentation are available from: http://wiki.cs.cf.ac.uk/twiki/bin/view/Sandbox/Ope nGridForum21 Grid community could make highly useful contributions in all of these areas. 3.14.1 EDINA had been a member of the OGC since 2000 and used GridNet2 resources to part-sponsor an OGC Technical Committee meeting at the University of Edinburgh during June 2006. For more details, see: http://edina.ac.uk/ogcconference/ This meeting had several ramifications for the UK academic sector and wider members of the geospatial community, but in terms of the stated GridNet2 objectives, the following were probably the most significant outcomes: • Mr Higgins used his GridNet2 funding to concentrate on developing links and liaising between the two main SDOs active in defining open interoperability standards in the Grid and Geospatial communities, respectively. • Open Grid Forum (OGF) • Open Geospatial Consortium (OGC) The OGC (http://www.opengeospatial.org/) is an international industry consortium established in 1994. It consists of approximately 300 companies, government agencies and universities participating in a consensus process to develop publicly available interface specifications. OGC specifications support interoperable solutions that “geo-enable” the Web, wireless and locationbased services, and mainstream IT. Grid-enabled Spatial Data Infrastructure A key concept necessary for understanding the context of much of this liaison work was that of Spatial Data Infrastructure (SDI). In recent years, SDI had become a principal driver in the development of Geographic Information (GI) standards and could assist in understanding where the GI and Grid communities intersected. The definition used by the Infrastructure for Spatial Information in Europe (INSPIRE) directive (entered into force May 2007) is as follows: SDI encompasses the policies, organisational remits, data, technologies, standard delivery mechanism and financial and human resources necessary to ensure that those working with spatial data, whether at the global or local scale, are not impeded in meeting their objectives. (INSPIRE consultation paper, 2003) Significant resources were being channelled into SDI development at the global, regional (European) and national level. Major challenges included interoperability, security and scalability, noting that many geospatial operations were characterised by being computationally intensive. It became apparent throughout the duration of this initiative that the OGC Committee Meeting – June 06 An ad hoc Grid GIS meeting (also open to non-members of the OGC) was held on 28th June, with the following speakers: • Malcolm Atkinson (National e-Science Centre) on GGF, OGSA, European Grid activities, NeSC • Pier Giorgio Marchetti (European Space Agency) on The Grid model in the ESA Ground Segment: status and perspectives. • Andrew Woolf (NERC Data Grid) on Wrappers, portlets, resource-orientation and OGC in earth system Grids • Neil Chue-Hong (National e-Science Centre) on Mapping OGC standards to OGSA-DAI: approaches and issues • A lunch meeting was organised between senior members of both communities to establish grounds for a working relationship. • The GridNet2 logo was prominently displayed as an event sponsor throughout the week. The report from this meeting, presented at the Technical Committee meeting plenary is available at: http://edina.ac.uk/ogcconference/gridgis.html 3.14.2 GGF18 This was the first GGF meeting that Mr Higgins had attended. Mr Higgins attended this meeting to familiarise himself with the structure of the OGF and with the full spectrum of Grid activities, and to conduct in networking. Mr Higgins met with key individuals in the Grid community, discussed the OGF position in respect of geospatial information and laid much of the foundation for the OGC OGF workshop at OGF22. Mr Higgins attended a large number of sessions, in particular: • Several sessions focusing on security • Earth Observation and Ground systems in distributed environments This session confirmed that the Grid community, in respect of geospatial information, had focused most of its attention to date on remote sensed data. This was unsurprising as the volumes of data generated were substantial and the processing of such data was consequently computationally intensive. This session also Page 87 3 GridNet2 Reports 3.14 Mr Chris Higgins – Edinburgh confirmed that there already was awareness of the work of the OGC and that there was no formal liaison to date. Importantly though, there was an appetite for establishing a formal link. Methodologies for Environmental Analysis (IMAA) The workshop concluded with a panel discussion. Mr Higgins made key Grid contacts at this meeting, and found it interesting to contrast the operation of the OGF with the OGC. The OGF came across as a larger organisation with a wider spread of members, including many very big players. The gathering was larger but less focused than within OGC. Characterised by larger academic sector involvement than OGC and much of the work discussed appeared to be closer to the research rather than the production end of the spectrum. The following questions and issues were raised during the presentations and discussion. 3.14.3 • OGF20 • An OGF-OGC workshop was held at OGF20. For more details, see: http://www.ogf.org/gf/event_schedule/index.php?id =707 The stated purpose of this workshop was to “…plan concrete collaboration between the Open Geospatial Consortium (OGC) and the Open Grid Forum (OGF) to identify and produce standard, Grid-enabled geospatial information tools in the context of a service-oriented architecture”. There were presentations from key stakeholders and potential adopters, followed by a panel session to get rough consensus on specific activities, such as harmonising the OGC Reference Model with emerging web services. A Memorandum of Understanding between OGC and OGF was also discussed.” The event had three components. The first section attempted to communicate to the Grid community the breadth, applicability, maturity and uptake of the open geospatial interoperability standards, noting that the latter included standards from ISO TC/211 as well OGC (these two organisations liaise very closely). • Introduction, Craig Lee, OGF Area Director, The Aerospace Corporation • OGC/ISO TC211 Standards Landscape, David Arctur, OGC Interoperability Institute • GEOSS (Global Earth Observation System of Systems), Jeremy Morley, University College London • United Nations Spatial Data Infrastructure (UNSDI), Kristin Stock, University of Nottingham, Social Change Online The second section concentrated on projects that were demonstrably bridging geospatial and Grid standards communities: • • • • There is a need to “migrate data”. • This could mean secure data transfer, thirdparty data transfer. There is a need for "real-time data processing”. • Distributed resource mgmt, scheduling, workflow management. • How does this relate to the WPS? How well does SAW-GEO meet most needs? • • • • • • • OGC: Digital Rights Management, Info Comm & Semantics groups? • OGF: Semantic Grid, Workflow Research Groups? Geospatial Interoperability • Data semantics, geo-service semantics, ontologies. • How does this drive geo-workflow capabilities? Geospatial reasoning SEE-GEO project, Chris Higgins, EDINA, Edinburgh University • SAW-GEO project, Gobe Hobona, University of Newcastle Activities at George Mason University, Liping Di, George Mason University Cyclops/EGEE, Stefano Nativi, Italian National Research Council (CNR), Institute of Page 88 General-purpose inference engines, planning agents? Web 2.0 vs. traditional Grid tools • • Dynamic service provisioning. Which groups in the two organisations might benefit from collaboration? • • How well does GT4 serve SAW-GEO’s needs? What does SAW-GEO need for wider examination, adoption? When would it be appropriate for geospatial community? Geo data virtualisation • “Recipes” for deriving data products from raw data. • Support for file/data replicas. Grid OGC Collision programme: • • • What does "Grid-enabling" OGC tools (WCS, WMS, WCS...) mean? HPC for geospatial data • Scalability? Outcomes The following outcomes resulted from the OGC workshop discussions: • Start new Geospatial Community Group (CG) in OGF. 3 GridNet2 Reports 3.14 Mr Chris Higgins – Edinburgh • • This group would be the point of coordination for further specific collaborative activities between OGC and OGF, for example, interoperability testing of OGC tools and data sets across Grid infrastructures. OGC-OGF Collaboration Workshop Major issues: In late 2007, the Open Geospatial Consortium and the Open Grid Forum signed a Memorandum of Understanding (MOU) on the topic of Grid-enabling standard geospatial processing tools. That MOU identified a set of possible technical topics to pursue. The motivation for this workshop was to bring together researchers and industrial practitioners working on those topics, which are: • Handling massive data. • Need organisations that can provide compute resources that can support the vast interoperability requirements (for example, TeraGrid, NaReGI, EGEE). • Develop joint interoperability specifications for geospatial processing services, workflow, and security. Mr Higgins was co-organiser of the OGC-OGF Collaboration Workshop, together with Craig A. Lee and Satoshi Sekiguchi. • The initial CG deliverable is to be an OGF Informational Document defining more specific collaboration efforts and opportunities. 3.14.4 OGF21 This could include generic web services, Web 2.0 mash-ups, and also complete, distributed service architectures with Grid security models. • Apart from attending the wide variety of sessions covering a variety of subjects and further networking, Mr Higgins’ two main reasons for attending this event were: http://wiki.cs.cf.ac.uk/twiki/bin/view/Sandbox/ OpenGridForum21 2. To participate in the OGC-OGF workshop, which Mr Higgins assisted in arranging. 3.14.5 OGF22 Mr Higgins attended OGF22 to continue engagement in activities which develop synergies between the e-Science/Grid and Geographic Information (GI) communities. This report complements presentations given at the OGC-OGF Collaboration workshop at OGF22. Copies of these presentations are available from: http://www.ogf.org/gf/event_schedule/index.php?e vent_id=9 In brief, as planned, two EDINA staff (Mr Higgins and Michael Koutroumpas) attended OGF22 and attended a number of sessions, the most significant being the all day OGC-OGF Collaboration workshop. UK e-Science Ramifications from OGF22 There was much discussion between UK and German representatives both at the workshop and afterwards. As a direct consequence of this workshop, plans were advanced for a 3-4 month collaboration between the UK SEEGEO and the German GDI-Grid projects. If successful, the demonstrator created would show geospatial resources being securely shared using standards based approaches across national e-infrastructures with consequences for both the JISC and wider UK e-Science. Integration of WPS with workflow management tools. Workflows could be managed by scripting languages such as Swift, by compiled code such as SAGA, or by workflow management engines such as DAGMan, Pegasus, Kepler, Taverna, and Triana, just to name a few. 1. To give a presentation at the GridNet2 session. For more details, see: Integration of OGC's Web Processing Service (WPS) with a range of back-end processing environments. • Integration of OGC Federated Catalogues / Data Repositories with Grid data movement tools and Grid security models. Data consumed and produced by WPS calls might come from and be returned to OGC Federated Catalogues in a simple client-server fashion, but secure, third-party transfers might also be desirable. These MOU items formed the point of departure. The real goal of this workshop was to get motivated people together, identify what people really need to do, get efforts snowballing between meetings, and to continuously engage key stakeholders and potential adopters. During the course of the workshop, Mr Higgins gave a presentation on SEE-GEO. This workshop was originally scheduled for two sessions but the response to the call for participation was so positive that a third session was added. The overwhelming response was primarily to due the number of on-going projects around Grids and geospatial systems. Specifically, the number of Gridrelated WPS implementations was surprising. Besides having a number of talks around Gridenabled OGC tools, there were also two talks concerning the fifth OGC Web Services test bed (OWS-5) and also OGC's approach to geospatial metadata and cataloguing services. The first session began with these two talks as a way of getting OGF folks more exposed to OGC methods. The OGC test bed efforts are an important mechanism for driving the development of best practices, standards and tools. Likewise, every application domain was most likely to have developed their own metadata schemas, ontologies and so on that distributed infrastructures (like Grids) would have to support. Clearly information registries and discovery services for both geospatial data and computing resources needed to co-exist and cooperate. Page 89 3 GridNet2 Reports 3.14 Mr Chris Higgins – Edinburgh The last talk of the first session on 52º North WPS-G was presented by Bastian Baranski of the Institute for Geoinformatics at the University of Münster. WPS-G features a pluggable framework that can support different Grid middleware, has full GML2 support for complex values, and supports both synchronous and asynchronous processing. The current implementation is UNICORE-based. WPS-G was demonstrated with an avi screenshot movie. • The entire second session comprised a set of talks from the GDI-Grid (Geodateninfrastrucktur-Grid) project, part of the German D-Grid project. D-Grid had the overall objective of producing a national Grid infrastructure with support for three major Grid middleware packages (UNICORE, gLite, and Globus) and, even more importantly, a self-sustaining business model. Hence, GDI-Grid had the same objectives in the area of geospatial applications, such as flood simulation, noise propagation, and emergency routing. It was important to note that while deeply entwined with D-Grid, this set of talks was led by lat/lon (www.lat-lon.de), a small company specializing in new technologies for distributed geospatial data management and processing. • These talks looked at using WPS as a Grid front-end with different translator services, design options for chaining multiple WPS calls (for example, workflow), and how to integrate Grid security mechanisms into OGC tools, such as WPS, in a way that is still easy and simple for non-Grid specialists to use. Continuing on the topic of Grid security, the third session had a talk by Higgins on SEE-GEO and the use of the Shibboleth Grid security system to provide authentication (single sign-on) and authorisation (digital rights management). A talk from ESRI (www.esri.com) was also originally scheduled for the first session but this was cancelled at the last minute and replaced by the talk from Traverse Technologies. The third session originally had a second talk from NASA Ames but this was cancelled shortly before the workshop. Nonetheless, the extensive discussion of these talks overflowed to the point where there was only about 15 minutes of remaining time for discussion at the end of the third session. The following are further references for the OGC Web Processing Service (WPS): • 52north.org/index.php?option=com_project s&task=showProject&id=21&Itemid=127 • Workshop Results and Future Work As already indicated, there was a very high response to the Call for Participation for this workshop and there were more Grid-enabled implementations of WPS than anticipated. This indicated a large interest in and motivation for precisely for this kind of tooling. This was surprising since geospatial data was applicable across many different application domains. A key challenge was how to hide the complexity of Grids and to make the end-user tools as easy to use as possible by non-specialists in geospatially-related application domains. If WPS became the accepted way (dominant practice) that people in the OGC community used for their distributed processing needs, then this might very well be the way to enable that large community to adopt Grid technology. There were a number of follow-on action items from this workshop: • OGF would set-up a project on GridForge where anybody can get documents, browse the wiki and so on, and people with a GridForge account could post items to the project. • Relevant groups in OGF would be encouraged to review the WPS 1.0 spec. These groups included the HPCBP-WG, the GRIDRPC-WG, and the SAGA WG, to better determine the breadth of supportability with Grid tooling. • OGC and OGF agreed to coordinate their meetings being held June 2-6 in Europe. OGF23 was in Barcelona, Spain, while the OGC Technical Committee meeting wase in Potsdam, Germany. Videoconferencing was explored as an option, along with coordinating the meeting schedules to allow interested parties to travel between the two cities. OGF23 was co-organised by OGF-Europe and was expected to have workshops on digital repositories, with participation by geospatialrelated projects such as DEGREE (www.eudegree.eu) and GENESI-DR (www.genesi-dr.eu). Other candidate technical topics for this coordinated meeting included security, workflow, and programming models. • The European Geophysical Union planned to have sessions on Earth & Space Science Informatics with several sessions explicitly covering Grids: ESSI8, ESSI9, and also ESSI10. European partners in the OGC-OGF collaboration might use this meeting as an opportunity to make further progress. Information for all ESSI sessions can be found at: The Approved WPS Standard, V1.0 WPS references this standard: OGC 06-121r3, OpenGIS® Web Services Common Specification Example uses of WPS: • Discussions, findings, and use of WPS in OWS-4 portal.opengeospatial.org/files/?artifact_id= 19424 • OWS-4 Workflow report - use of WPS in workflows portal.opengeospatial.org/files/?artifact_id= 19778 • Organisations that publicly announced WPS code sets. Page 90 PyWPS (Python Web Processing Service) pywps.wald.intevation.org www.opengeospatial.org/standards/common • lat lon wiki.deegree.org/deegreeWiki/WebProcessin gService www.opengeospatial.org/standards/wps • 52 North http://www.cosis.net/members/meetings/progra mme/session_programme.php?p_id=323&PHPS 3 GridNet2 Reports 3.15 Mr Neil Chue Hong – OMII-UK ESSID=5c977de80d06eea2e1ac4f2c24f9ae2c&P HPSESSID=5c977de80d06eea2e1ac4f2c24f9ae2c • • • • • • Grids in the IT Data Centre • Town Hall • DAIS Working Group Session • OGSA-DAI Workshop • Standards All-Hands Meeting • caGrid 1.0 – Update on caGrid Infrastructure Project • OGSA Data Architecture • Data Integration Solutions with OGSA-DAI • OMII-Europe • Data Area Meeting • SAGA Session INFOD was also discussed as OGC was examining event-based approaches, for example, pub/sub. There was interest in reviewing an INFOD presentation on sensors that could benefit from using OGC Sensor Web Enablement standards. • OGC/OGF Collaboration Workshop (2 sessions) • Software Providers meet GIN and Standards • OMII-UK: Reducing the gap between researchers and resources (2 sessions) Vladimir Getov (Westminster, UK), one of the editors for IEEE Computer, invited OGC and OGF to co-author a paper on this collaboration, to be targeted for the November 2008 issue, just in time for SC08. Relevance OGC and OGF would explore possibilities for concrete engagement in OWS-6 that was in the planning process. As a result of this workshop at OGF22, the German GDI-Grid and UK SEE-GEO projects held discussions about further collaborations. Both projects were examining issues relating to providing access to geospatial information on national Grid infrastructures and there was significant overlap, particularly in the area of security. Virtualisation was mentioned several times in the workshop and OGC had a sidebar discussion about the implementation of virtualisation in the GEO-Grid project (AIST, Japan). This was an opportunity as OGC is leading a related task in GEO. 3.15 Mr Neil Chue Hong – OMII-UK Mr Neil Chue Hong was appointed director of OMII-UK after the departure of Dr Steven Newhouse. Mr Chue Hong inherited Dr Newhouse’s GridNet2 award, which he used to fund OMII-UK attendance at OGF meetings, in keeping with the original purpose of the award. For a summary of Mr Chue Hong’s GridNet2 activites, see section 2.15, Mr Neil Chue Hong – OMII-UK. 3.15.1 OGF21 Mr Chue Hong attended OGF21 in Seattle and this report describes the relevance of the trip to current OMII-UK and UK e-Science activities. Mr Chue Hong attended the following meetings: • Opening Session • OGSA-ByteIO Interoperability Fiesta results • Web 2.0 Grids and Cyberinfrastructure • Real-time Web 2.0: Evolution of Middleware for Grid-based Instruments and Sensors • ByteIO Experience Document compilation • OGF Marketing Workshop • Keynote: Think Little: The Proliferation of Small Clusters Means Big Changes • Standard API for Data Grids Mr Chue Hong was primarily interested in the application of standards to the creation of sustainable, interoperable implementations of software of use to researchers. In particular, he contributed directly to many of the data area standards groups. Mr Chue Hong co-chaired the ByteIO-WG, and this was a significant OGF for that group, because they presented the results of the ByteIO Interoperation Fiesta, which compared the four implementations of the standard. The group also started work on the experiences document. Mr Chue Hong attended a number of sessions concerned with interoperability which featured software which had been developed or sponsored by OMII-UK. He also participated in a panel Software Providers meet GIN and Standards on this subject. Mr Chue Hong participated in a number of the discussions in the more mature working groups to understand the current status of the specifications and their routes to adoption, for instance through OMII-UK software – this included DAIS-WG, and SAGA-WG, as well as the work done through OMIIEurope. Mr Chue Hong chaired the OMII-UK workshop, which had a good attendance comprising appoximately 25% of the total OGF attendees. There were demonstrations of a number of the software components developed by OMII-UK and their relevance to the OGF community. This was a good opportunity for OMII-UK to obtain better feedback, and it also improved the visibility of OMII-UK’s work and the UK e-Science projects associated with this work. Mr Chue Hong participated in the OGF/OGC Collision workshop, which featured work done by OMII-UK in the UK with EDINA, Leeds and Manchester as part of the SEE-GEO project. It also featured work being done using OGSA-DAI, an OMII- Page 91 3 GridNet2 Reports 3.16 Dr Richard Hughes-Jones – Manchester UK software component, in the GeoGRID project led by AIST in Japan. draft, and Mr Chue Hong solicited comments on the document from a number of other OGF WG chairs. Mr Chue Hong had a number of discussions with different OGF attendees about OMII-UK and their efforts to deliver and support software communities. Mr Chue Hong also talked to people working on UNICORE and Globus, and continued discussions about the OMII-UK/caBIG collaborations. Mr Chue Hong attended a number of sessions concerned with interoperability which featured software which has been developed or sponsored by OMII-UK. Mr Chue Hong pressed for prioritising available effort on getting interoperability at the critical points for allowing people to migrate between different infrastructures. As NomCom Chair, Mr Chue Hong attempted to solicit volunteers for the community led nominations process. He participated in the Marketing workshop to give a UK perspective. Overall, by being able to attend OGF21, Mr Chue Hong was able to promote a lot of the standards and e-Science related work that OMII-UK are doing or sponsoring. He was also able to get a good perspective of what work, particularly in standards, would be relevant for UK e-Science software, and thus for OMII-UK. 3.15.2 OGF22 Mr Chue Hong attended OGF22, and participated in the following sessions: • Opening Session • OGF Marketing Workshop • OMII-Europe: Using open standards to deliver interoperability • OGSA-DMI Discussion of Specification • OGSA-Data Architecture Future Directions • Authz Interoperation Demonstrations • Activity instance document schema • HPCP Specification Adoption • Grid Interoperation Now: Specification Adoptions & Discussions • SAGA: The Simple API for Grid Applications • SAGA + DAIS: next steps • Creating a standard software API for Data Grid Management Systems BoF • The Encyclopedia of Life: A Web page for every species • Financial Services workshop • DAIS-WG Session • Data Management Workshop Relevance Mr Chue Hong was primarily interested in the application of standards to the creation of sustainable, interoperable implementations of software of use to researchers. In particular, he contributed directly to many of the data area standards groups. Mr Chue Hong was co-chair of the OGSA-ByteIO WG which did not have a session at this OGF, but which came up in the context of many discussions. The OGSA-ByteIO-WG had an experiences document in Page 92 Mr Chue Hong participated in a number of the discussions in the more mature working groups to understand the current status of the specifications and their routes to adoption, for instance through OMII-UK software. The groups included DAIS, SAGA, and Authz. Mr Chue Hong had several discussions with Shantenu Jha and Andre Merzky of SAGA about socialising and promoting their work, and how to tie it into the data access and data transfer specs. Although unable to attend the WFM-WG, Mr Chue Hong had discussions with Ian Taylor and contributed input based on the experiences of OMIIUK collaborators and users. Likewise, though he was unable to attend the OGC-OGF collision workshop, to which he has contributed at past OGFs, Mr Chue Hong had a discussion with one of the OGC-OGF cochairs about its status and how OMII-UK can continue contributing. Mr Chue Hong had a number of discussions with different OGF attendees about OMII-UK and their efforts to deliver and support software communities. He also talked to people working on UNICORE and Globus, and continued discussions about the OMIIUK/caBIG collaborations. As NomCom Chair, Mr Chue Hong had discussions about raising the grassroots involvement in the steering and coordination of OGF. He also had discussions about strategies for improving the feedback cycle between the standards and e-Science councils, and continued his participation in the OGF Marketing group. Overall, by being able to attend OGF22, Mr Chue Hong was able to efficiently discuss a number of collaborations and generate new work because of the critical mass of people attending. He was also able to get a good perspective of what work, particularly in standards, will be relevant for UK e-Science software, and thus for OMII-UK. 3.16 Dr Richard HughesJones – Manchester Dr Richard Hughes-Jones is the Senior Experimental Officer in the School of Physics and Astronomy at the University of Manchester. For a summary of Dr Hughes-Jones’ GridNet2 activities, see section 2.16, Dr Richard Hughes-Jones – Manchester. 3.16.1 GGF15 The main focus for Dr Hughes-Jones at GGF15 was the NM-WG. Dr Hughes-Jones was one of the cochairs at GGF15, and so organised a full day working meeting, an open meeting and a tutorial on Web Services for NM Applications. The attendance at the meetings was good with 10 – 15 people at each 3 GridNet2 Reports 3.16 Dr Richard Hughes-Jones – Manchester session, and many of these were NM-WG “core” people. NM-WG Sessions • IPv6 and multicast • Validation Authority for Schemata It was agreed to publicise a set of valid schemata for tools in current use. This would be ranked by number of users. It needed to allow extensibility, because there would be new schemata as new tools develop. Presentations Jason Zurawski presented the current status of the V2 schemas, which are available as a tarball on the NM-WG web page. New items included: European Update • IDs & IDrefs – allows to track state • Filtering Mark Leese gave a presentation of the current status of European NM projects. • Characteristics Software Demonstrations • Parameter block There were demonstrations for the following NM software: V2 Schemata Discussions The following V2 schemata issues were discussed: • The item may be in the metadata section, but then one would have to give all the metadata each time any ONE item changes, which causes an XML explosion problem. There was also a need to be able to group measurements together depending on other things, such as the interface number and not just the time. • perfSONAR – Jason Zurawski Jason gave a demo of PerfSONAR. This was a way to use the V2 schema to exchange historical SNMP counter data. It was being tested in Dante and Internet2 networks. • EGEE-JRA4 & European Update – Mark Leese Subsequent Discussions • The session that was open to all was a summary of Monday’s meetings regarding the group’s progress and so on. • Web Services pseudo-tutorial: Web Services for NM Applications • Need examples • Traceroute Marlon Pierce started by giving an introduction to Web Services and then described the following in more detail: • Stats from an interface • WSDL 1.1 overview Approaches actions: • SOAP 1.2 overview • Don’t normalise – just zip. • SOAP formats • Use of pre-defined format – and use this as a handle. • Data encoding • Explore use of RDF to help this XML explosion. • Message routing • Seek input from web service community for transmission of repeated metadata. Units Jason led this discussion. SI units were agreed. A discussion took place about how to express units values and it was agreed there should be a units attribute. • Performance Advisor – John Estabrook Schema normalisation Jeff Boot led the discussion. There may be groups of related items and only one or two things change, such as a lot of data with same characteristic, and only the time changes. • • Response Codes The current thinking was covered in the slides for this meeting. The NM-WG planned to ask the people involved in WS. It would be clearer if <characteristic> was replaced by <response>. If care was taken with choosing the namespaces URI, it should be able to be mapped onto URLs which could be useful for the future. Shrideep Pallickara then gave a presentation and led the discussion on advanced capabilities of Web Services including: • Addressing • Eventing • Reliability In summary, at this GGF meeting, the NM-WG made significant progress in discussing and resolving many of the technical issues involving the Version 2 NM-WG schemata that had been brought up on the mailing list and phone calls. Dr Hughes-Jones believes the presentations on WS clarified many issues and helped the group. Page 93 3 GridNet2 Reports 3.16 Dr Richard Hughes-Jones – Manchester 3.16.2 OGF19 Dr Hughes-Jones attended various working group and research group sessions at OGF19, and also went to the OGF town hall meeting. Astronomy Applications Group Discussions Network Measurement Working Group Sessions Dr Hughes-Jones chaired all the NM-WG sessions at OGF19. An immediate aim was to contact other OGF working groups to ascertain whether the needs of Astronomy analysis were being catered for by the Grid community. The first session included a summary of the OGF.23 Draft Standard and a review of the current position of the version 2 Schemata. It was agreed to progress a draft standard document on the version 2 Schemata, check for any minor updates to the OGF.23 Draft Standard, and then produce an OGF Experimental document to move OGF.23 to a standard. The Astro RG has created a requirements document (OGF informational document). The second session worked on the document for the version 2 Schemata. The needs of the Astro RG included focus on using specialised “workbench” (portal) techniques to perform global searches of their data archives for sky images. Unlike particle physics, the astronomers did not tend to write analysis programs or code for a given investigation. Grids and System Virtualisation Group In response to a question, the group was not (yet) considering the needs of Radio Astronomy, which was using 512 Mbit/s real-time data flows from the telescopes to the correlator, and researching how to make use of Grid farms for distributed correlation. Storage Networking Community Group After the meeting, Dr Hughes-Jones had discussions with the co-chair about participation in the e-Science stands planned for OGF20 in Manchester. OGF Opening Tuesday 30 Jan It was interesting to note the emphasis on tangible deliverables and the emphasis on producing draft standards. It was also good to see the improving alignment of academic and industrial/commercial interests. Firewall Issues Research Group The FI-RG meeting was a very good meeting with many informative presentations. The different firewall techniques were presented and discussed in detail. The focus was on the security of setting up switched Lambdas or Lightpaths between computing sites/hosts. For example, if user A was correctly authenticated by the middleware and set up a Lightpath between two sites, how was user B, who was not allowed to use this link, prevented from making use of the Lightpath? This was very relevant to the current Remote Farm work by the ATLAS Trigger/DAQ group and others. Particle and Nuclear Physics Applications There was a good discussion of needs and issues, during which it was suggested that contact with the Astronomy Applications Group would be beneficial. Infrastructure Area Meeting This was a small meeting. It reviewed the current work of this area and discussed whether investigation of other infrastructure items like storage would be useful. It appears that the groups currently in the Infrastructure Area are progressing well. It was agreed to continue examining other areas that might be included in the infrastructure area. Page 94 This group will examine the OGF work to see if the use of virtual machines and systems needs extra support of features to operate in the Grid. There was a presentation of the proposed work on storage management in the Grid from two storage vendors. They focused on commercial data centres, but the issues and requirements seemed very similar to the work that has been going on in the academic community Grids and in the OGF Standards Council. There seemed to be no plan to create use-case requirements or best practice documents but rather to ask the group to produce “code” to solve the perceived problems in commercial data centres. Grid High Performance Networking Meeting There were several presentations on different techniques being investigated for network resource management and Lightpath or link reservation. There were questions and discussion of the issues of interoperation between different Grids when considering network resource management. There was also much discussion on the new draft of the Grid User Network Interface document, which seems to be making progress. 3.16.3 OGF20 Dr Hughes-Jones was very impressed by the knowledge and enthusiasm of the young members of EGEE who manned the stands in Manchester Central. In spite of the expense, he thought it was an opportunity well taken to put the European and UK e-Science work on view to the world Grid community. Dr Hughes-Jones attended various sessions at OGF20. GFSG Dr Hughes-Jones attended his first steering group meeting as Area Director for Infrastructure held on Sunday afternoon 6th May. Infrastructure Area Meeting The meeting was held at 14:00 on Monday 7th May. As AD, Dr Hughes-Jones co-chaired this small 3 GridNet2 Reports 3.16 Dr Richard Hughes-Jones – Manchester meeting with Franco Travostino. The purpose of this meeting was to review the groups currently in the Infrastructure Area to determine progress, concerns and potential problems. The groups reviewed were: • Grid High-Performance Networking RG (GHPN-RG) • Network Mark-up Language WG (NML-WG) • Network Measurements WG (NM-WG) • Grid and Virtualization WG (GRIDVIRT-WG) Grids and System Virtualisation Group It was agreed that NM-WG and NML-WG would work on a method to translate between the existing network description schemata. EU Keynote Presentation Mario Campolargo gave a very useful update and forward look at the EU PF7 calls and road map. Firewall Issues Research Group The RG documents were discussed; the first is already provided as GFD-I083 and it was proposed to merge the other two that are in preparation into one document. Initial thoughts on the future of the RG were discussed. The meeting was held at 18:00 on Monday 7th May. This was the second meeting of the WG and the two co-chairs gave presentations that covered the following: Grid High Performance Networking Meeting • Ideas for use cases • Update on GUNI (Georgios Zervas) • Concept of a virtual workspace • • Management of virtual machines in a Grid infrastructure Enlightened Computing Update: An HD-class example (Jon MacLaren) • SIP protocol for Grid Networks (Aldo Campi) GridHypervisor project • Research Challenges for Optical Grid Networks (Marc De Leenheer) • Introducing EC-GIN: Europe-China Grid InterNetworking (Sven Hessler) • OGF Opening & Keynote The keynote speech for OGF20 was given by Tony Hey on The Social Grid and examined the challenges in defining standards to make Grids interoperable. The presentations at this meeting covered: GridNet2 Meeting Network Measurement Working Group Sessions Dr Hughes-Jones attended the GridNet2 Advisory Board meeting at the University of Manchester. Mark Leese & Dr Hughes-Jones co-chaired all the NM-WG sessions at OGF 20. 3.16.4 The first session presented updates on the NM-WG work and Alexander Ploss from Muster University gave a talk on Network and Runtime Monitoring in the edutain@grid SOA. The other sessions continued work on the document for the version 2 schemata. Network Markup Language Working Group. The NML-WG meeting covered input for the informational document in the following presentations: • Introduction to NDL, the Network Description Language from Amsterdam • Schema of the Network Measurements WG showing the need for network description • Model vs. Syntax and a review of the draft document on XML vs. RDF • Vision of RDF descriptions of Grid resources, such as databases, files, network, CPUs, visualisation, and so on There was an invitation to everyone to provide requirements for the schemas. OGF21 Dr Hughes-Jones attended various meetings, working group and research group sessions at OGF21. GFSG Craig Lee of The Aerospace Corporation introduced himself as the new OGF President and outlined a road map of future directions for OGF. The items presented in the opening meeting and at the Town Hall Meeting were discussed. The meeting continued in break-out sessions. Chris Smith led discussion in the Standards Function. One of the main issues was to set up a process to review the health and activity of all the working and research groups in the Standards Function. This was to include: • Emphasis on use of the OGF mailing lists • Updating the living charter so that the work was relevant to Grids & OGF • Focusing the work on the charter • Ensuring that the documents were produced There was a proposal to remove the research groups from the Standards Function. While some research groups may have finished their work; for others, this was not a problem and did not need to be fixed. Page 95 3 GridNet2 Reports 3.17 Dr Jens Jensen – STFC OGF Opening Craig Lee introduced himself as the new OGF president and gave a very clear road map of future directions for OGF. He outlined OGF's goal to build participation in the research/academic community by: • Promoting peer-reviewed publications related to OGF work • Co-locating future OGF events with established research Grid conferences • Collaborating closely with other research organisations such as the Open Geospatial Consortium • Grid High-Performance Networking RG (GHPN-RG) status is green. • Network Mark-up Language Working Group (NML-WG) status is green. • Network Measurements Working Group (NM-WG) status is green. • Grid and Virtualization Working Group (GRIDVIRT-WG) status is red/orange (see above). Franco, the other AD, was to shepherd the GRIDVIRT-WG. Standards All Hands Meeting Network Mark-up Language Working Group. Chris Smith led the discussion of the work in the Standards Function. The NML-WG held two sessions at OGF21. Grid High Performance Networking Meeting • At session 1, there were presentations describing current work being done on network modelling. This will act as input for the first deliverable, which is an OGF informational document describing the context of the NML work. • At session 2, the discussion focused on the issues that should be included in the first schema. Network Measurements Working Group Dr Hughes-Jones chaired all three sessions of the NM-WG. The first session covered status and update reports. Mark Leese from joined by telephone and gave an update of what was happening in Europe with EGEE-II and in the UK with Gridmon that monitors the networks used by the UK e-Science and Particle Physics communities. Hughes-Jones presented the work done by the GEANT2 JRA in extending the schemata to include end2end status information about LightPaths. The other two sessions were spent in detailed discussion of the Draft Recommendation Document specifying the schemata. It is worth noting that the work of this group has been taken up by 29 networks world wide – one of OGF’s successes. Grids and System Virtualisation Group Both co-chairs of this group attended the session by telephone, and their presentations seemed more like an introduction, rather than proposals for interfaces in virtual machines or system virtualisation concepts that could be standardised by an OGF WG. The co-chairs are active but there are few members of the working group. Infrastructure Area Meeting As AD, Dr Hughes-Jones chaired this small meeting. After communicating the operational suggestions from the GFSG, the meeting reviewed the groups currently in the Infrastructure Area to determine progress, concerns and potential problems. The groups and status are: Page 96 This meeting was chaired by Cees de Laat. There was a review of all the documents from the group, and all seem to be making progress. There were presentations on the following: • Status of GNI • Progress on Network Research • Report from the GLIF meeting Firewall Issues Research Group After some presentations on the current work, there was discussion on whether to put the RG on hold and form a working group to start on a specification/work around the "virtualisation" of firewalls. GridNet2 e-Science Workshop Members of the UK presented summaries of their work on Grid issues and at OGF. Dr Hughes-Jones gave a talk on the work being done on networking by the WG and RG in the Infrastructure area. 3.17 Dr Jens Jensen – STFC Dr Jens Jensen works in Data Services at the STFC’s Rutherford Appleton Laboratory. For a summary of Dr Jensen’s GridNet2 activities, see section 2.17, Dr Jens Jensen – STFC. 3.17.1 OGF22 Dr Jensen had three main purposes for attending OGF22: 1. To represent the Storage Resource Manager (SRM) protocol collaboration (GSM-WG), as chair, to advance the protocol as an OGF standard. 2. To present advances and issues regarding operations of Grid Certification Authorities (CAs) in the CAOPS-WG. 3. To propose new standards for CA operations, in the International Grid Trust Federation (IGTF). 3 GridNet2 Reports 3.17 Dr Jens Jensen – STFC The IGTF defines standards and minimal requirements for Grid CAs, but also meets in CAOPS-WG even if it doesn’t cover operational issues. GSM-WG storage systems from the Grid but still permitted extremely efficient access to them. He also included an overview of how the SRM met biomedical use cases in EGEE. Finally, Dr Jensen briefly discussed seven areas of particular challenges facing the SRM protocol. Dr Jensen chaired the two Grid Storage Management (GSM-WG) sessions. The aim of these sessions was to review the current SRM version 2.2 specification, which is the current OGF standard. The problem was there were six interoperating implementations (with a seventh on the way, see GIN below), which have all been implemented and have had interoperations issues resolved since the document was written. As with any standard, the process of implementing it highlighted areas where the specification was not sufficiently precise. Moreover, one of the major customers, the Worldwide Large Hadron Collider Computing Grid (WLCG) imposed requirements upon the implementers which deviated significantly from the original specification. Thus, anyone following the specification as-is would not be able to interoperate. The slides are available at the following location, although they are designed to accompany the talk, not stand alone: Since Dr Jensen was leading this effort, he gave the only presentation in this session, but there were high quality discussions (mostly technical) with members of the audience. Dr Jensen’s talk, however, contained input from at least three of the six implementers, as well as issues from the WLCG deployment, so Dr Jensen was not just speaking for himself. CAOPS-WG There were about a dozen people present, slightly different people between the two sessions. For the benefit of newcomers, Dr Jensen gave a 20-30 min introduction to SRM, which was an abridged version of his talk for the data management track (see below). Then Dr Jensen went on to the technical talk, highlighting issues with 2.2 and discussing options for further progress with members of the audience. It was intentional to keep this session technical to ensure that the standardisation discussion is open, but also because other data management or storage groups may benefit from the technical issues which concern GSM-WG. Although the talks were advertised as two independent talks, Dr Jensen ended up combining it into a single set of slides. There were 30 slides (including the title) and they took more than two hours to get through. http://www.ogf.org/OGF22/materials/1143/datamg mt.ppt GSM-WG/GLUE-WG Previously, Dr Jensen has contributed to the GLUE 1.3 and 2.0 work to help define the Storage Element in the GLUE schema. Unfortunately the GLUE Storage Element session clashed with a CAOPS session in which Dr Jensen was speaking, and the GLUE people were unable to reschedule, so apart from minor discussions in the coffee breaks, Dr Jensen was not able to contribute. In this CAOPS-WG session, which lasted a whole day but also contained the IGTF sessions, Dr Jensen presented the following talks: • This was a talk about protecting private keys for X.509 certificates, particularly for services which depend on the availability of certificates themselves, such as VOMS services and CAs themselves. The talk focused on the conflict between having long-term recoverable copies (curation) and having protected copies (confidentiality). The aim of this talk was to gather recommendations for CAs in particular to do this. The talk contained technical recommendations. During the talk, there were suggestions from members of the audience which added to the body of knowledge. The outcome of this should be a recommendations document and a small software repository. • http://www.ogf.org/OGF22/materials/1052/SRM22issues.ppt The introductory talk was a subset of the talk Dr Jensen gave in data management the following day: http://www.ogf.org/OGF22/materials/1143/datamg mt.ppt The outcome of this should be a whole separate document of recommendations. Dr Jensen was asked by Erwin Laure to present data management issues on behalf of GSM-WG in Thursday’s data management track, which unfortunately clashed with the CAOPS track. Dr Jensen’s talk introduced the SRM protocol, explaining briefly why it is different from the Storage Resource Broker (SRB). In particular, Dr Jensen emphasised the role of the Storage Element, and how the SRM met the conflicting goals of protecting Certificate renewal The purpose of this talk was to ask whether certificates can be renewed, which is where an X.509 certificate is re-signed by a CA using the existing key pair. This was useful only if clients were able to keep their private keys and replace the old certificate with a new one – a process which was trivial for some types of clients, but in other cases needed testing or even additional tools to help the process. In this case, feedback based on the combined experience of the audience suggested that support for renewal is well understood. The technical talk is available here: GSM-WG/Data Management Private key protection • CP/CPS model template In this talk, Dr Jensen presented the aims of a working group which he is leading. The aim was to shorten the process for accreditation for a new CA wishing to become a member of the IGTF. In turn, IGTF membership implied that the CA would be trusted as a Grid CA by some of the largest Grids in the world, such as WLCG, OpenScienceGrid, EGEE, as well as most national Grids, including TeraGrid and the NGS. The aim, in brief, was to make it easier to write an IGTF- Page 97 3 GridNet2 Reports 3.17 Dr Jens Jensen – STFC compliant Certificate Policy and Certification Practices Statement by introducing intelligent re-use (as opposed to the current occasionally non-intelligent copying and pasting from existing CP/CPSs). Dr Jensen knew about work in TAGPMA on providing guidelines, which was helpful for the non-template parts, but he learned at the OGF that NAREGI were also looking into doing something along the line of what he wants for templates. IGTF In these sessions, which identify minimal requirements for Grid CAs, Dr Jensen gave the following talks: • http://www.ogf.org/gf/event_schedule/index.ph p?id=1062 http://www.ogf.org/gf/event_schedule/index.ph p?id=1127 http://www.ogf.org/gf/event_schedule/index.ph p?id=1129 Additional Notes • Dr Jensen was contacted by the chairs of Grid Resource Allocation Agreement Protocol (GRAAP-WG) and Alexander Papaspyrou from TU Dortmund. They were interested in scheduling storage resources. Dr Jensen mentioned old discussions among GSM-WG regarding scheduling networking resources between Storage Elements for special transfers. There may be opportunities for future discussions and collaborations, and they had already exchanged email following this contact. • Wearing an NGS hat, Dr Jensen needed to follow up with NGS regarding resource usage records for RUS-WG. • Still wearing the NGS hat, Dr Jensen talked to D-Grid (FZK) about the JISC-funded SARoNGS project and promised to send details. He also discussed SARoNGS and existing input projects with Professors David Chadwick (Kent) and Richard Sinnott (Glasgow). • Dr Jensen contributed to Grid Interoperability Now (GIN) activities (currently mainly GINDATA with recent demonstrations of SRM and SRB interoperability). Dr Jensen promised in the GIN session to review the updated version of a paper for the special issue of J. Grid Comp. Robot certificates “Robots” are automated clients, software agents which can act on behalf of a user but without users being present. Examples include monitoring systems which regularly run test jobs, automated data movers, and portals which run jobs on behalf of the user (that is, the user is anonymised by the portal). The UK e-Science CA was the first Grid CA to implement robots (although this is not obvious as the software doesn’t support it directly and the RAs have not been trained to support it), at least outside the US, and it aimed for a cautious implementation (to avoid relying parties deciding they don’t trust us), an implementation which was subsequently copied by other CAs (so far, the Netherlands and Italy, with Czech Republic to follow). However, the UK e-Science implementation was significantly different (that is, more “cautious”) from what the Americans had implemented trying to solve the same problems. This talk aimed at presenting the (European) implementations, which had also differed among themselves in some respects, to see if a common minimal requirements list for robots could be found. Clearly existing implementations couldbe documented – this talk does that – but agreement was still some way off. The Americans felt the European model is too cautious. • http://www.ogf.org/gf/event_schedule/index.ph p?id=1056 Moreover, in GIN, it was announced that OMII-Europe might be able to contribute effort – “can take over some of the more tedious tasks” – which would be most welcome. GIN also needed to involve more production Grid people and standards people. Higher Level CAs This wasn’t a talk, as such, but the session looked at a document which Dr Jensen had written, but with input already from several contributors. The group previously identified a need for requirements for CAs that were needed on the Grid to build trust chains to a self-signed root CA, but which would not issue certificates to people or hosts. In this session, Dr Jensen quickly went through version 0.3 of the document, which has previously been discussed only in a narrower group. The aim was to make it an IGTF standard for all the Policy Management Authorities (PMAs) in the world (these are the bodies which review and accredit Grid CAs). Dr Jensen explained some examples, and was actioned to add the examples to the document (which he hadn’t previously for lack of time) and circulate it again. This document is now available in GridForge, under the CAOPS document repository. The following sessions discussions are available at: Page 98 • Dr Jensen talked to a few people about security issues – in particular, authentication – in medical research. Dr Jensen obtained some contacts for people (also industry) with relevant experiences, and some who were interested in my experiences. He made additional notes where some talks (for example, Catlett’s) had suggestions. • Four different Hardware Signing Module (HSM) vendors were expected to attend the Thursday morning CAOPS session, although, unfortunately, one didn’t make it due to bad weather at his airport. This was relevant because many CAs use HSMs in various forms, from the high-end, used by the UK e-Science CA, to key protection tokens used in some (not yet in production) NGS-specific credential conversion CAs, which aim to broaden the NGS user base by abstracting certificate management away from the users. Also (European) robots require hardware tokens. Dr Jensen followed up with a major vendor of tokens, who promised to find out from their UK branch information regarding pricing for NGS 3 GridNet2 Reports 3.18 Dr Shantenu Jha – UCL use as the token prices have been prohibitively expensive in the past. Dr Jensen was later contacted by their UK representative. The Brazilian government HSM project (Kryptus) is also interesting, as they would almost certainly be cheaper than many existing vendors. Dr Jensen knew about the project already but had an opportunity to talk with their representative. • • • • • • Dr Jensen managed to attend all three of Reagan’s sessions, more by accident than design. Catching up with Arun in the coffee break was probably more useful, as Reagan mostly gives the sales pitch. Nevertheless, since the UK has a number of SRB services and a few proposed iRODS projects, it was worth keeping track of the developments, particularly the migration strategy from SRB to iRODS. Dr Jensen did learn that iRODS 1.0 had been released and managed to download the source. Moreover, comparing NARA’s work with “data Grids” to Dr Jensen’s in curation and long term digital archiving was fairly useful. The number of projects using workflow engines, in particular, Taverna, was increasing. This was also the case for astronomy communities, for whom Dr Jensen’s group provides storage services, who are interested in interoperability. For file management, a number of minor issues cropped up which were worth keeping an eye on (perhaps in GIN-DATA). In the Grid File System (GFS-WG) session, this included creative uses of the LFC file catalogue and comparing it to RNS, and mounting the latter in Windows. There was work being done to make RNS accessible from gLite. There wasn’t much information about storage in GFS this time. Dr Jensen discussed access management again with the TeraGrid folks (specifically NCSA and NERSC via DoEScienceGrids). This was useful for two reasons: firstly, because it is necessary for TeraGrid to trust the UK e-Science CA for the purposes of interoperation, and also because Dr Jensen was looking at looser models of authentication and assurance in the UK, and also for the NGS. NCSA mentioned NanoHub had a “peered hosts” use case fow which there was no solution yet found (as the obvious solutions would violate minimal requirements for certificate management). The CMS experiment (part of WLCG) had a similar problem (if Dr Jensen understood the NanoHub problem correctly) with PhEDEx/FTS services in the UK. Dr Jensen missed the XAM part of the data management session because he was giving talks in CAOPS. It would have been useful because (with his storage service provider hat) such technology potentially could enable us to scale storage services to the next level, for example, by switching to object based storage (OSD). The idea was that some complexity and functionality is pushed down closer to the storage device, enabling data servers to manage more systems. XAM is the evolving standard access protocol for this in SNIA. project manager), and to prepare his presentations. • On Friday, there was meant to be a GIN activity meeting but nobody seemed to be around, Dr Jensen took that day off (claiming no expenses), spending the day at MIT with Marcel Kunze from FZK (head of e-Science at IWR). 3.18 Dr Shantenu Jha – UCL Dr Shantenu Jha was a Visiting Fellow at the Centre for Computational Science at UCL. For a summary of Dr Jha’s GridNet2 activities, see section 2.18, Dr Shantenu Jha – UCL. 3.18.1 GGF15 For Dr Jha, GGF15 was primarily about re-chartering the SAGA-RG and working out the details of SAGARG spawning a SAGA WG. Both missions were accomplished. Technical details of the three SAGA-RG meetings are available at: • Meeting No. 1: http://www.ggf.org/mail archive/sagarg/2005/10/msg00001.html • Meeting No. 2: http://www.ggf.org/mail archive/sagarg/2005/10/msg00008.html • Meeting No. 3: http://www.ggf.org/mail archive/sagarg/2005/10/msg00009.html The main accomplishments from the meetings were: • The final touches were added to the use case document and it was submitted to the GGF editorial pipeline. For more details, see: http://www.ggf.org/mail archive/sagarg/2005/10/msg00006.html • There was a meeting of the SAGA officers with the Grid Forum Steering Group to discuss “bit flipping” and the alignment of SAGA with SOA in general, and OGSA in particular. The notes for the meeting about bit flipping are available at: http://www.ggf.org/mail archive/sagarg/2005/10/msg00017.html • There were many technical details and issues discussed, and many of these issues were resolved. • Dr Jha delivered a 10 minute update on SAGA at the closing plenary. Dr Jensen stayed Sunday night (that is, overnight Sat-Sun) to reduce the cost of the flight by over 50%. Dr Jensen took advantage of this day, using it to discuss storage and project management with Patrick Fuhrmann from DESY (dCache Page 99 3 GridNet2 Reports 3.18 Dr Shantenu Jha – UCL 3.18.2 GGF16 For Dr Jha, GGF16 was primarily about: • Finishing the re-chartering process by finalising and passing the charter for the newly spawned SAGA-CORE-WG. • Discussing the issue list for the SAGA Strawman API. • Discussing the SAGA requirements document – authored by Dr Jha and Andre Merzky The technical details of the four SAGA-RG meetings are available at: • http://www.ggf.org/mail archive/sagarg/2006/05/msg00023.html • • Discussing issues of compatibility between ongoing (nascent) implementations of SAGA. The last SAGA-RG session was devoted to a discussion of this important issue and brief presentations by the 3 main implementation groups. Technical details of the four SAGA-RG meetings are available at: • Meeting No. 1: http://www.ggf.org/mail archive/sagarg/2006/02/msg00103.html • Meeting No. 2: http://www.ggf.org/mail archive/sagarg/2006/02/msg00104.html • Meeting No. 3: http://www.ggf.org/mail archive/sagarg/2006/02/msg00105.html • Meeting No. 4: http://www.ggf.org/mail archive/sagarg/2006/02/msg00106.html The other main details are as follows: • Discussions began about SAGA-RG and GridRPC group coordination • Discussions began about GIN and SAGA involvement. • Discussions began about SAGA-RG and GridCPR and a timeline was set. 3.18.3 GGF17 For Dr Jha, GGF17 was primarily about working through the crucial, but unspectacular and unromantic, technical details. Other important achievements in the immediate runup to GGF17 were the submission of the requirement document to the editorial process and the publication of the use case document as GFD.70. See: http://www.ogf.org/documents/GFD.70.pdf Page 100 Meeting No. 2: http://www.ggf.org/mail archive/sagarg/2006/05/msg00024.html • The bulk of the work for the requirements document was done at the design team meeting at LSU in December 2005. The requirements document had been circulated on the SAGA-RG list and to the GGF roughly 15 days in advance; the public discussion was a sanity check before completion and submission. Meeting No. 1: Meeting No. 3: http://www.ggf.org/mail archive/sagarg/2006/05/msg00041.html • Meeting No. 4: http://www.ggf.org/mail archive/sagarg/2006/05/msg00042.html 3.18.4 GGF18 At GGF18, the SAGA group had two sessions. Both of these sessions included presentations to update the group on technical developments on the API and discussions. Although the sessions were primarily concerned with the API, some specific use cases were discussed as well as some outreach activity. In a way GGF18 was a meeting where there was discussion of the API as part of the process of converging towards a stable uniform first release aimed for early 2007. Dr Jha, as co-chair of the SAGA group, set the agenda for and chaired both meetings, and moderated the discussions. For more details about the meetings, see: http://www.ogf.org/pipermail/saga-rg/2006September/000888.html 3.18.5 LGC Workshop The Lightweight Grid Computing (LGC) Workshop took place at Castleton in the Peak District, UK, in May 2006. A description of the workshop is available at: http://tyne.dl.ac.uk/GROWL/Lightweight.shtml Dr Jha was asked to give a talk on SAGA and how it might relate to middleware being developed in the UK. The main aim of this workshop was to share ideas on the middleware needed to provide users (read applications) with lightweight access to Grid resources. In a way, this was a functional level below the level at which SAGA operates, but it was crucial that middleware developers understood the kind of abstractions and functionality required of applications. Although this event was primarily a SAGA dissemination event for the UK e-Science community (of which Dr Jha anticipates a few more), there was useful feedback from several projects/groups to SAGA. The single most important feedback for the SAGA-RG was the need for programmatic interfaces for “client toolkits which can provide very light-weight but extensible access to Grid resources”, not to mention job and resource management. 3 GridNet2 Reports 3.19 Mr Jipu Jiang – NeSC 3.18.6 SAGA Design and Implementation Meeting The SAGA Design and Implementation Meeting was held January 2006. Dr Jha met with Andre Merzky (Vrije University), Harmut Kaiser (LSU), who were both also very active in the SAGA group, to work primarily on implementation discussion (progress and technical issues) and complete a first draft of the SAGA requirements document. This was an important goal to achieve before the next GGF meeting, GGF16. This was a meeting away from the glare and distractions of the GGF, and it was held in a secluded environment enabling the three attendees to focus on the low-level details of the SAGA Strawman API specifications, to clean up things, to create an issue list and most importantly, prepare a draft of the SAGA requirements document so it was ready for distribution. The group was successful on all these accounts. 3.18.7 ISSGC 07 This was the first and major effort at SAGA outreach. With the final specification very close to submission to the OGF Editorial and GFSG Board, the invitation to speak at the International Summer School on Grid Computing to about 80 enthusiastic students was a good opportunity to “test” various aspects of the specification. Additionally the ISSGC provided the opportunity to discuss many aspects of SAGA with Prof Miron Livny and others. This has led directly to a project with him to investigate SAGA as a programmatic interface to Condor system. The talk and material presented are available at: • http://www.iceageeu.org/issgc07/sessionDescription.cfm?id=132 3.19 Mr Jipu Jiang – NeSC Mr Jipu Jiang was funded by Professor Richard Sinnott’s GridNet2 award. Mr Jipu Jiang is a Grid engineer at NeSC at the University of Glasgow. For a summary of Mr Jiang’s GridNet2 activities, see section 2.19, Mr Jipu Jiang – NeSC. 3.19.1 GGF18 As a member of NeSC and researcher on the JISCfunded Glasgow implementation of Shibboleth (GLASS) project, Mr Jiang attended the GGF18 conference held in Washington DC, US. Mr Jiang attended the following sessions: • GGF18 Opening • Keynote: Vision for 21st Century Discovery • Shibboleth for Grids: Experiences and Interoperability (all four sessions) • Security Talk Session • OGSA-Authz-WG • Education and Training Community Group Workshop • GGF Closing • Storage Grids in Healthcare • Topics in Identity Management GGF18 Meeting Details Mr Jiang’s primary reason for attending GGF18 was for the security-related work, particularly the Shibboleth for Grids workshop, which lasted for two days. Mr Jiang’s work in NeSC is mainly focused on Grid security. The GLASS (GLASgow early Adoption of Shibboleth) project, on which Mr Jiang was working, was designed to explore and to implement Shibboleth technologies in the University of Glasgow. Updates since the Grid and Shib BoF at GGF16 in Athens were presented at this workshop. A talk was given by Prof Richard Sinnott on behalf of NeSC to introduce the new work being done in both GLASS and DyVOSE project for adopting Shibboleth technology in the Grid environment. Through attendance at this workshop, Mr Jiang gained a better understanding of Shibboleth-related work being done internationally, and related components, such as new versions of Shibboleth. Mr Jiang found that the most interesting talk in the Shibboleth for Grid workshop was on the MAMS project from Australia. The group have done a lot of integration work between Shibboleth and the GridSphere Portal framework. They now have a whole set of tools running on the GridSphere portal framework. Although not much of their work involves Grid, they did some impressive explorations in portal and Shibboleth integration. Mr Jiang believed that future collaboration with them could greatly improve the work done at the NeSC. To expand and update his knowledge in the Grid Security field, Mr Jiang attended the Security Talk session held immediately after the fourth Shibboleth for Grid workshop. The delegate, Blair Dillaway from Microsoft, presented an XML-based Assertion Language for security policies called SecPAL, and gave an interesting demo after the talk. Mr Jiang attended the Education and Training Community Group Workshop. He expected that this workshop would be of use in teaching the Advanced MSc students, which is one of his duties in NeSC. As one of the aims of the DyVOSE project, a Grid Computing module was created to provide hands-on Grid training to Advanced Computing Science MSc students at the University of Glasgow. The methods learned from the workshop would be very helpful in delivering the course materials to the students. Besides adopting Shibboleth to the University of Glasgow, the GLASS project was also going to use this technology to provide secure access to NHS resources/data. Through collaboration with Glasgow Southern General Hospital, a fine-grained portal is planned to be built to provide access and usage of data from brain trauma patients. To gain better understanding about the Grid technologies in the health domain, Mr Jiang attended the Storage Grids in Healthcare session. Mr Jiang went to the Topics in Identity Management session in the last timeslot of the GGF18. Understanding issues raised in identity management from other people’s experiences was crucial to a success adoption of Shibboleth to the unified account management system in University of Glasgow. Page 101 3 GridNet2 Reports 3.20 Dr Mike Jones – Manchester To conclude, Mr Jiang found GGF18 to be a wonderful and rewarding opportunity which allowed him to present NeSC-related work and understand other people’s work and experience at the same time. Mr Jiang hoped to continue these efforts in the future. 3.20 Dr Mike Jones – Manchester • Work was in progress to shibbolise MyProxy and allow it to run as an on-line CA. • Most projects were addressing the AA in the SOA space. Overall the UK was well represented, contributing more than half of the presentations. Dr Mike Jones is a Grid Support Engineer in the Research Computing Services at the University of Manchester. In the GIN-CG meetings, volunteers were sought from the different Grids. The NGS was put forward as a contributor and Dr Jones was volunteered to represent NGS on matters relating to Authentication and Authorisation. For a summary of Dr Jones’ GridNet2 activities, see section 2.20, Dr Mike Jones – Manchester. Dr Jones also participated in other sessions at GGF16. 3.20.1 3.20.2 GGF16 GGF17 GGF16 Athens was Dr Jones’ first GGF meeting for some time. His main purposes for attending this meeting were as follows: There were no sessions on Shibboleth and the Grid at the GGF17. Dr Jones attended sessions focused on authentication and authorisation. • Attending the GridShib Investigators meetings as he is involved in a project to help the NGS to leverage Shibboleth. Dr Jones attended the following parallel sessions: • OGSA Authz New Protocols • Discussing field authentication authorisation issues relating to NGS within the GIN-CG sessions. • GT4 Status & Experiences, Applications and Deployments • Security Area session • Re-engaging in the community • GIN Progress and Plans • FI-RG • CA-OPS Session • TeraGrid Security: Managing security across a Grid Dr Jones attended the following parallel sessions: • GIN Meetings • OGSA Authz • Grid and Shib: Investigators meetings • IGTF • CA-OPS – OCSP, Revocation of proxies, Namespace constraints • Ad Hoc BoF on Grid applications of virtualisation technologies • Interoperability Fests • Grid Authorization – Interoperability Here & Now This proved to be an interesting session, especially the handling of the compromise on the TeraGrid. • Workshop on Data Access and Integration (tourism) Most of Dr Jones’ contributions were outside the scope of formal meetings at GGF17. Instead, he participated in informal discussions and networking on the topics of identity management and Grid security. Dr Jones also obtained information about the following: Dr Jones attended the FI-RG where he suggested the addition of AccessGrid as an example of UDP and multicast network traffic for the Firewall Issues Document. He also asked for the document to include the effects of source IP filtering and suggested that a section on Network Address Translation be added to highlight the plight of compute resources whose batch nodes are either firewalls or not connected to the WAN. All three suggestions were embraced and published in GFD.83. • 3.20.3 Dr Jones presented work on SHEBANGS and also the integration of Shibboleth into GridSite at the Grid and Shib meetings. SHEBANGS was well received, although the general impression was that it was quite complex due to the attempt to explain the technical aspects of the architecture. Dr Jones followed this up with those who raised questions during the meeting. SAML 2.0 had new features such as Single Logout, Attribute Encryption and the Enhanced Client or Proxy. • A final release of Shibboleth 2.0 was expected by the end of summer 2006. • Autograph provided dynamic configuration of Shibboleth attribute release. Page 102 GGF18 The main purpose of Dr Jones’ attendance at GGF18 was to attend the four Shibboleth and Grid experiences workshops. Dr Jones attended the following parallel sessions: • Shibboleth for Grids: Experiences and Interoperability 3 GridNet2 Reports 3.20 Dr Mike Jones – Manchester There was another long discussion about DN namespaces and the decision to drop OCSP. • OGF101 • CA-OPS • OGSA Authz WG • International Grid Trust Federation • FI-RG At the LoA-RG, the group identified two documents to be produced: • Aggregating Mobile Devices with Grids • • Topics in Identity Management A risk analysis in relation to LoA and use case gathering in an e-Science context • A gap analysis of current LoA definitions versus LoA requirements in e-Science/Grid context • Most of this GGF meeting was taken up by Shibboleth activities. Dr Jones presented the current status of SHEBANGS, highlighting the problems regarding identity credential namespace, the method by which the Levels of Assurance (LoA) are conveyed, and some experiences with the configuration of Shibboleth. gLite LoA-RG The group also discussed various authentication scenarios and identified the areas relevant to LoA. The work on LoA was further propagated to the Grid community through the circulation of the ES-LoA questionnaire in various relevant sessions. Of particular note were the developments of Shibboleth 2.0 (still not released) which appeared to remove the WAYF from the federation. Most of the inner workings seemed to be majorly different to the Shibboleth 1.3 architecture in order to embrace IdP discovery and single sign-out. David Chadwick presented what looks to be a useful policy editing application that allows service providers to express complex policies in natural language using XLST transforms. Outside of meetings, Dr Jones was involved in various discussions. In particular, Dr Jones had a discussion with David Groep, David Chadwick and Frank Siebenlist, regarding current and future authorisation mechanics in Globus based Grids, which specifically covered the ability to include UID and GID assertions in authorisation engines such as LCMAPS and GUMS. Groep also highlighted issues relating to the existence of GIDs and UIDs on back end nodes. Two documents were discussed at the CA-OPS meeting: Unfortunately, Dr Jones was unable to make it to the WHISKY which was rescheduled. • Guidelines for Authentication Service Profiles for Grids 3.20.5 • Grid Certificate Profile The main purpose of Dr Jones’ attendance at OGF21 was to participate in the LoA RG, and security sessions in general. There were long and quite heated discussions about what a federation means within this group, identity management groups, the wider Grid community and the on-line community in general. The need for a glossary document was suggested. 3.20.4 OGF20 The main purpose of Dr Jones’ attendance at OGF20 was to participate in the LoA BoF and to man the NGS and Manchester booths. Dr Jones attended the following parallel sessions: OGF21 Dr Jones attended the following parallel sessions: • Web 2.0 – Grids sessions • OGSA Authz • Security Area Meeting • Standard API for Data Grids (GFS-WG) • Storage Resource Managers – practical experiences (GSM-WG) • OGSA-Authz (OGSA-AUTHZ-WG) • Grids and Server Virtualization • LoA-RG • CAOPS Session (CAOPS-WG) • Joint Session on Information Modelling for Computing Resources (GLUE-WG) • caGrid 1.0 Update on caGrid Infrastructure Project • JSDL General (JSDL-WG) • SAGA – Software Solution Session • JSDL 1.0 Extensions (JSDL-WG). • IGTF Session (CAOPS-WG) For Dr Jones, it was unfortunate that this session was co-scheduled with the Workshop on Heuristics for Implementing Semantic Knowledge Yardsticks (WHISKY). • JSDL Working Session (JSDLWG) • FI-RG • LoA Session (LoA-RG) • GIN • Dynamic Service Level Agreements • CA-OPS Session (CAOPS-WG) Due to session scheduling issues, Dr Jones was unable to attend the GIN sessions and the majority of the GridNet2 e-Science sessions. Page 103 3 GridNet2 Reports 3.20 Dr Mike Jones – Manchester OGSA Authz WG Dr Jones and Richard Sinnott were the designated note takers, and the notes are available here: topic) about X.509 DNs and SOAs. Globus (WS) were to implement signing policies at the beginning of 2008 In summary, the two existing specifications were reviewed: It was a surprise to Dr Jones that there were no signing policy checks in GT4 WS. He thought that he may have missed the details, but after a quick search through GT4 Java code base, he found no signing policy or EACL strings. • XACML Profile FI-RG The XACML Profile now has a good level of uptake by middleware vendors. The main thrust of this session was the proposal of a group to take forward the idea of a web service authorisation framework for the dynamic opening of ports. Both Dr Jones and Richard Hughes-Jones noted that there were similarities between current SAML callout work from the authZ group that might have a natural synergy to this. http://forge.ogf.org/sf/go/doc14890 • WS Trust Profile Two new specifications were proposed: • • SAML Attribute Retrieval Profile to cover the query to VO's AAs for membership information retrieval SAML VO Attribute Profile to cover the VO, Group and Role attributes It was noted that the security documents were generally one step ahead of implementations, which was seen to be a good thing. However, this caused a void in user community feedback, blamed in part on the lack of research funding into this area. It was also observed that, in general, the security documents were now becoming too complex for general reading (This would be OK if all applications, vendors, and virtual organisations had their own security experts; this might not be the case). Security Area Meeting Mike Jones (Microsoft namesake, not the subject of this report) gave an interesting presentation on MS CardSpace. It seems to have a nice GUI layer with backend plug-ins into most authN/Z systems. There was a promise of open standards and quite a few similarities with other projects in the same space, for example, Shibboleth and Autograph. GSM-WG This session was again a presentation-oriented session rather than a dialogue with the community. There appear to be 5 interoperating SRM implementations: LoA-RG Dr Jones presented the findings from the recent JISC funded ESLoA project. These were well received as the foundations for the first document A Gap Analysis of Current LoA Definitions versus LoA Requirements in e-Science/Grid Context. Jones stressed that although the community at large seemed to identify LoA and risk-based access control as a requirement, the community had responded in such a way as to suggest that the topic was not well understood; there was a chasm in understanding between Grid and non-Grid subscription-based services. The second document from this group A Risk Analysis in Relation to LoA and Use Case Gathering in an e-Science Context has yet to gain traction, as the security modelling within e-Science was not sufficient at this time to obtain a large enough pool of use cases. Dr Jones attended the other groups as a session tourist. 3.20.6 OGF22 The main purpose of Dr Jones’ attendance at OGF22 was to participate in Security Area meetings, to attend the RUS and UR WGs, and to attend the OGC sessions. Dr Jones attended the following parallel sessions: • RUS-WG • BeStMan • OGSA-Authz-WG • CASTOR • Security Area Meeting • DPM • Authz interoperation demos • dCache • OGC-OGF Collaboration workshops • StoRM • GSM-WG sessions CAOPS-WG • UR-WG The Grid Certificate Profile had now passed through the public comments phase, the comments were to be pushed to WG final call 06/11/07, and then the document was to be passed onto the editorial committee. • CAOPS – IGTF Workshops The current version of the Audit Guidelines document was presented, and a synergy was noted between this document and the auditing processes for the TAGPMA. Namespaces were discussed and the usual suspects raised their concerns (favourite Page 104 RUS-WG Morris Riedel chaired the session and was happy to see that NGS was represented. Before the session started, Dr Jones and Riedel discussed why the NGS had interest in this area, and it was well received that NGS was using RUS as a component in its accounting model. Dr Jones believed that the group wishes to 3 GridNet2 Reports 3.21 Mr Ian Kelley – Cardiff maintain ties with the NGS to receive further feedback. During the session, other RUS implementations were discussed: • Unicore6.1 implementation • DGAS gLite Accounting System • SGAS Globus incubator project All of these projects planned to have a release at the end of February 2008. They were aware of the JISC Accounting call and wished to form ties with any activity funded in that area. Dr Jones gave a brief description of how the RUS was implemented within the NGS. There was further discussion regarding RUS, the GLUE schema and the CIM. OGSA-Authz-WG (This includes the Authz Interoperation Demonstrations.) There was some discussion about the architecture document and whether components needed to be restated in the profile documents. The group was just about ready to pass the Attribute Exchange Profile to area directors. The XACML profile (only 7 pages) was converging with that of EGEE/OGF's, but it was, however, envisaged that they would not merge into one document due to time constraints on the latter. Worries about having two documents meant that the group was relying on a good public comment phase. The WSTrust document had only one implementation so far, from David Chadwick's group. There was a discussion about attribute and obligation standards, an area where it might be difficult to reach consensus. Security Area Meeting It was noted that there was no LoA RG meeting; activity had slowed down. Dr Jones mentioned that the Gap Analysis draft was removed from the GridForge site, due to the document being not to the LoA-RG Chair's preferred standard, and it was removed at her request. Dr Jones pointed out that there had been much work on revamping the document and that he would ask for the document to be put back on GridForge. OGC-OGF The OGC Workshop consisted of three meetings. Dr Jones attended these meetings to identify whether there were any crossovers between the SARoNGS project and security-related access control to OGC Web Service interfaces. There is an open source community called Deegree (http://www.deegree.org) that is providing OGC interfaces with an official OGC reference implementation. There was some interesting discussion about adding Meta GeoData to XACML policies (gXACML?) for access control. There was discussion about the possible extension to OGSA-DAI to directly query OGC Web Services and WPS using OGAS-DAI as a toolkit. Worries about the performance of OGSA-DAI were raised and it was mentioned that later versions of the middleware had improved performance significantly. Interesting presentations from SEE-GEO, D-Grid and EDINA led to further discussion on the formation of a Virtual Organisation and further collaboration between these 23 groups and the NGS. GSM-WG This group looked like it was becoming more standards oriented. However, it was noted that version 2 of the SRM specification was now out of date compared to all implementations and that anyone wishing to build an interoperable SRM had better engage the community instead of reading the developing specification. There was a large discussion about what was meant by 'space'; it seemed that different implementations dealt with space differently and so the behaviour could be predicted. UR-WG This meeting was very poorly attended as there were only five people, consisting of two chairs from this group, two chairs from other groups and Dr Jones. The group is looking for a new chair (from outside UK). UR is published but had many known shortcomings. For example, it was not good for noncomputational resources and it was unable to handle aggregate usage well. Dr Jones pointed out that VO membership was missing and that NGS was having to add this into the resources extensions. It seemed this was the only extensible part of the UR specification v1. There is a survey available at: http://forge.ogf.org/short/urwg/ Dr Jones assured the group that NGS would be feeding their experiences back through the survey. There was a brief overview of the JISC-funded Review of Grid Accounting & Usage report, followed by a good discussion between the attendees about how the Unicore and NGS systems worked. Future work on the UR v2 was then discussed. It was worrying about the level of complexity but perhaps this was a requirement. CAOPS-WG This was an all day session. It started with presentations from a number of HSM vendors. The representative from Aladdin, who produce the eToken, was questioned about the cost and openness of the drivers. He mentioned that there was a new driver released that unified the behaviour across Linux, Windows and Mac. There was some scepticism and it may only be the newer javacard versions of the eToken that behave consistently, not the Siemens version. Jensen presented the rest of the talks, of which the one that was the most thought-provoking was on Robot Certificates where a debate about naming was entered into: Why must CN=Robot:... and the answer was that this was due to the logging facilities. One further interesting discussion came from another talk: The opening of the IGTF OID space for proxy certificates to contain host information for debugging. Dr Jones mentioned that this was a big missing part in the current Proxy certificate security model, and would partly address what to do if a machine in the field with valid proxy certificate and keys was compromised. 3.21 Mr Ian Kelley – Cardiff Mr Ian Kelley is a PhD student in Distributed Collaborative Computing at the School of Computer Science, University of Cardiff. For a summary of Mr Kelley’s GridNet2 activities, see section 2.21, Mr Ian Kelley – Cardiff. Page 105 3 GridNet2 Reports 3.21 Mr Ian Kelley – Cardiff 3.21.1 GGF18 Mr Kelley met with Michael Russell, the developer of Grid-Sphere’s GridPortlets, at GGF18 and talked at length with him about the proposal, and got his support. Russell explained in detail the current architecture of his new Vine project that will replace GridPortlets, and how it helps to solve some of the given problems. Mr Kelley used this as an opportunity to give him the ideas outlined in the proposal, and he seemed keen to architect his new system to further support the needed interoperability. Mr Kelley and Russell both went to attend the scheduled GCE-RG meeting, which was unfortunately cancelled but still on the OGF agenda. However, they met a developer from Indiana there, and had a good discussion with him, which included these issues. Mr Kelley successfully gained the support of other GridSphere developers at SC06 (Novotny, Wehrens). OGF19 would be used to further build up the community and promote the issues and find adequate solutions. 3.21.2 OGF20 Sharing Workflows Mr Kelley participated in the Sharing Workflows session at OGF20. Although the general session focused on the need for sharing workflows for scientific applications, there was a presentation for an OMII-UK portals project in which Mr Kelley is involved. This is the WHIP project. The WHIP project is seeking to build standardised portal mechanisms for accessing shared artefacts, or digital objects, for example, workflow descriptions. OGF could prove a very good forum to get more involvement in this project from other communities, especially other OMII-UK institutions, and make the work more standardised and useful for external projects. An overview of the WHIP project was presented by Andrew Harrison (Cardiff University), and discussions followed offline with other OMII-UK groups to synchronise efforts on some projects and make the different software interoperable. Other presentations that are interesting from the portals perspective were: • myExperiment - Social Software for Workflow Sharing – David De Roure, University of Southampton and OMII-UK • Shibboleth Protection and Management of Workflows – Richard Sinnott, National e-Science Centre, University of Glasgow OGSA Workflow Mr Kelley attended this session to see the current status of OGSA workflow and potentially how it would relate to WHIP and other portal issues. The session was interesting, but rather unrelated to many of the portal issues. Service Oriented Knowledge Utilities (SOKU) Session This session was interesting to see some of the new technologies that are being adopted and brought forth in the Web 2.0 community. It was useful to Page 106 attend to see the possible future directions that portal development could take such as AJAX. Additional Sessions Mr Kelley also attended these sessions: • e-Arts and e-Humanities • Grid: a means to what end? • Scaling Up to the Enterprise Level • Collaborative Grids • Various keynote speeches These sessions were interesting to show the directions in which the Grid was moving, especially from the view of corporations that were deploying Grids, and the tools and technologies they were using in this facility. In addition, following a brief meeting with Rob Allan at OGF20 in which Mr Kelley and Allan discussed their related portal projects, Mr Kelley contacted Allan, Michael Russell (GridSphere), and Marlon Pierce (OGCE) regarding the status of the GCE-RG at OGF. Pierce, Russell and Mr Kelley had a discussion about the credential issues related to OGCE and GridPortlets, came up with some initial ideas, and agreed that the Web 2.0 sessions at OGF21 would provide a good opportunity for people to meet, share ideas, and find collaborations. 3.21.3 OGF21 Mr Kelley attended OGF21 in Seattle, which took place Oct 15 - 19, 2007. Mr Kelley attended various sessions, mostly the Web/Portals related sessions that were happening in the main theatre. Many interesting topics came up during these sessions, and it seemed as if the portal efforts might very well be re-invigorated with this new OGF shift towards collaborative environments, Grid computing environments, and Web 2.0 technologies. Following discussions after OGF20 between Michael Russell (GridSphere/PSNC), Mr Kelley, and Marlon Pierce (OGCE), Mr Kelley was able to organise a meeting with Michael to talk further about the more advanced portal interoperability issues and get a technical presentation of his upcoming software package, entitled Vine. Vine looks very interesting and has many prospects regarding the issues of portal back-end interoperability and development toolkits for building portal components. Additionally, Mr Kelley met with other Portal developers, including GridSphere’s Oliver Wehrens (AEI) and talked briefly with Gregor von Laszewski about the portal efforts he would be working towards in the near future. Beyond the portals discussions and meeting attendance, Mr Kelley was actively involved in the full-day GridNet2 meeting, where a colleague from Cardiff, Dr Matthew Shields, presented much of the work Cardiff University is undertaking within GridNet2. After initial presentations by many groups and individuals, much discussion took place concerning the prospect of continuing GridNet2 by submitting a new proposal. Mr Kelley was actively involved in this discussion, as well as in a discussion on how to improve communication between researchers and information exchange within the 3 GridNet2 Reports 3.22 Dr Amrey Krause – Edinburgh consortium. At the end of the meeting, when plans were being made on how to proceed further, Mr Kelley volunteered to help with any new proposal should it materialise. 3.21.4 OGF22 Mr Kelley participated primarily in the WFM-RG sessions and some additional sessions at OGF22. WFM-RG The general Workflow Sharing and Interoperability session focused on the need for sharing workflows for scientific applications. There was an interesting discussion on the responses to the questionnaire that was circulated before the session asking for use-cases and practical guidelines for the sharing of workflows. Mr Kelley participated in this discussion and asked some questions regarding the different application groups and the scope in which workflow-sharing was envisioned (between application domains, or solely within domains, etc). There was a committee developed to take the results of this questionnaire and write them up in an OGF document. Additional Sessions Mr Kelley also attended several additional sessions throughout the conference and had the opportunity to meet with GridSphere (and Vine) developer Michael Russell (PSNC) who gave him a demonstration of the new work he is involved in using Adobe Flex (Flash) to enable more interactive portal development (Web 2.0). This was very interesting and Mr Kelley thought that some of this work could be used within the WHIP project to enable the portal-side sharing of workflows. Flex also seemed to be a potential technology for interactively building and visualizing workflows within a web environment. It was decided that the best meeting for the initial EDGeS BoF would be at OGF23 in Barcelona, as many EDGeS partners would not be attending OGF22. (OGF22 was quite small, with only 240 participants.) Mr Kelley expected to be attending OGF23 on non-GridNet2 funds to pursue this aspect of his application. 3.22 Dr Amrey Krause – Edinburgh Dr Amrey Krause is an Applications Consultant at the Edinburgh Parallel Computing Centre. For a summary of Dr Krause’s GridNet2 activities, see section 2.22, Dr Amrey Krause – Edinburgh. 3.22.1 GGF16 Working Groups Data Area Meeting All data area groups reported on the current status of their work, including the DAIS-WG. DAIS-WG There were two DAIS-WG meetings. The following updates were presented to the community: • Norman Paton presented an overview of the WS-DAI specifications and the progress since GGF 15. • A new document has been added to GridForge: Interoperability Plan for DAIS Working Group Specifications, discussing how interoperability between DAIS implementations can be verified. • An implementation of WS-DAIX built on top of OGSA-DAI was demonstrated by Steven Lynden. The implementation highlighted several issues and comments on the WS-DAIX specification which would be addressed after GGF16. • RDF implementations of the WS-DAI specification were presented by groups from UPM (Madrid) and AIST (Japan). The DAIS charter would be amended to include an RDF realisation of the specification. The RDF implementers organised a F2F meeting taking place in Edinburgh after GGF16. Dr Krause had spoken to members of these groups. DAIS-WG Further Actions The groups involved in the RDF realisation arranged a F2F meeting in Edinburgh in June 2006. Dr Krause registered and planned to attend this event. Comments on WS-DAI received during the public comment period would be addressed by the editors. ByteIO-WG There was an update on the status of the ByteIO specifications. Michel Drescher demonstrated his implementation of the ByteIO specification. An interoperability meeting involving those groups who had done an implementation of ByteIO was agreed upon, although no specific dates were planned. Unfortunately, Mark Morgan, who was an author of the specification and had also provided an implementation, did not attend the meeting. During the last F2F (which Dr Krause did not attend) a quick interoperability test failed between Dr Krause’s implementation and Morgan’s. Byte I/O WG Further Actions An interoperability meeting between the implementers of the spec was planned. Dr Krause had written a simple implementation of ByteIO giving access to a file system and was planning to incorporate this into OGSA-DAI for delivery of BLOBs from a database or similar. Dr Krause hoped to have a working prototype of this by the interoperability meeting. There were some issues with interoperability with the implementation which Dr Krause would have to fix. Converging Web Services Standards BoF This was a meeting to inform the community about the convergence of the WSRF specifications and the WS Management specifications. There was a lively discussion about the convergence and migration issues. The OGSA-DAI team was concerned about this as they had to migrate to the merged standards eventually. This meeting was very useful as it gave an overview of what is going to happen and it made Dr Page 107 3 GridNet2 Reports 3.22 Dr Amrey Krause – Edinburgh Krause aware of the issues. The merger was not happening very soon, so it was more of a long-term issue to be aware of in general. Workshops The RDF realisation F2F meeting in Edinburgh in June was also discussed. ByteIO-WG OGSA-DAI Technology Update The ByteIO session gave an overview of the ByteIO group and specifications, and an update on progress including implementations and interoperability work. There was an OGSA-DAI workshop at which an overview of OGSA-DAI was presented and an update given on the features of the latest OGSA-DAI version. OGSA-DMI working group session There were some interesting discussions with current and prospective users of OGSA-DAI. Data Access and Integration This workshop had presentations from e-Science projects that were working on data access and integration problems, especially from Asia, on the current status of their work. This was interesting mainly from an OGSA-DAI point of view. OGSA-DAI was being used by many groups, especially in Asia. Meetings The OGSA-DMI session concentrated on the initial work on the functional specification. Workshops Mini-Symposium/OGSA-DAI User Forum The mini-symposium was an OGSA-DAI workshop at which an overview of OGSA-DAI was presented and an update on the features of the latest OGSA-DAI version. Globus Data Management for Architects Michel Drescher/ByteIO This workshop gave an overview of the data management tools available as part of the Globus Toolkit. Dr Krause had a follow-up to the ByteIO-WG meeting with Michel Drescher in which they discussed implementation issues. This was very useful. Education and Training Community Group Workshop The following WS-DAI specifications which Dr Krause ha co-authored were returned from the GGF editor after the public comment period: The Education & Training Community Group (ET-CG) planned to develop a programme of work to increase awareness of the importance of education and training for Grid computing, sharing best practice and developing standards for quality assurance and mechanisms for sharing E&T resources. • Web Services Data Access and Integration – The Core (WS-DAI) Specification Others • Web Services Data Access and Integration – The Relational Realization Progress • Web Services Data Access and Integration – The XML Realization Issues and comments regarding the ByteIO specifications were addressed by authors. Progress WS-DAI specifications Dr Krause also attended the OGF Town Hall. A response to the comments that were received during the public comment period was published on GridForge. 3.22.3 ByteIO comments DMI-WG Meetings Issues and comments regarding the ByteIO specifications had been submitted to the authors. There were two sessions for the DMI-WG which covered an overview and discussion of the functional specification. 3.22.2 GGF18 Working Groups Data Area Meeting All data area groups, including the DAIS-WG, reported on the current status of their work. DAIS-WG OGF20 Working Groups RM-WG The RM-WG sessions covered the scope and initial discussions for the group. SAGA Overview There was an overview of SAGA API and first demonstrations of SAGA API implementations. There was an update to the community on DAIS-WG activities and the group discussed progress towards the development of data access services for RDF data resources within the WS-DAI framework. DAIS WG Two specifications have been drafted for RDF data resource access. These were presented at the meeting. The interoperability testing status was discussed. Page 108 There was an update on progress and discussion on RDF-S specifications. 3 GridNet2 Reports 3.22 Dr Amrey Krause – Edinburgh There was a presentation and discussion on RDF-S use cases and specifications. Workshops Globus and Community This workshop gave an overview of the latest progress in the Globus community. One session concerned the results of the OGSA-ByteIO Interoperability Fiesta that was started 20 July 2007 were presented. The other session covered creating and compiling an Experience Document for OGSA-ByteIO. This mandatory document would be the last publication of the OGSA-ByteIO WG, advancing the ByteIO standard to an OGF Standards Recommendation. There were presentations on various database access tools on the Grid. An example for higher level interoperation between Genesis II and OGSA-DAI services was discussed. This might be taken forward as an MSc project in EPCC. OGSA-DAI users group meeting DAIS-WG There was an overview of OGSA-DAI and future plans were discussed. There were OGSA-DAI user presentations on the following projects: There was an update on progress and discussion on RDF-S specifications. Data management convenor OGSA-DAI users group meeting • Bridges Project (Richard Sinnott) • VOTES Project (Anthony Stell) • SBRN GEMEPS • Nano-CMOS • GEODE (Larry Tan) This workshop considered the complete spectrum of data work across OGF from the higher level vision being defined by the OGSA data architecture group; the data standards being produced such as DAIS and ByteIO; the technologies realising these standards such as OGSA-DAI; and then to the end user developers trying to building applications with these technologies. • SEE-GEO and GEESE (Chris Higgins) Workshops • VOMS on OGSA-DAI (Valerio Venturi) Globus and Community Meetings Dr Krause attended the following meetings: • OGSA-DAI and information retrieval (Greg Newby, University of Alaska Fairbanks) • Globus committers meeting • LIBI Project (EGEE2): OGSA-DAI Performance testing (Giacinto Donvito) Progress ByteIO interoperability testing There were four workshop sessions for Globus and Community. • The first session had a high-level overview, as well several in-depth presentations, for Globus tools. Dr Krause presented an overview of OGSADAI 3.0. There was a closing panel for community questions. • The second session discussed the growing complexity of the Globus toolkit and how this is being addressed by the Globus Toolkit team. • The third session covered standards, looking at the following questions: Public comments on the interoperability festival document have been received. • Are standards important for the components? Discussions of interoperability testing documents are ongoing. • Which standards are being used? An interoperability test was planned for July 2007. EPCC was planning to participate with an implementation of the ByteIO interfaces for OGSADAI. DAIS interoperability testing EPCC was planning to provide an implementation of WS-DAIX on OGSA-DAI. 3.22.4 OGF21 Working Groups • Dr Krause found this to be a useful workshop. Other Workshops Dr Krause also attended the following workshops: • GridNet2 Workshop • OMII-UK The OMII-UK workshop looked at reducing the gap between researchers and resources ByteIO-WG There were two sessions for the ByteIO-WG. The fourth session looked at the question of how Globus components work together. It would be useful to see projects that brought the Globus components together. • Data Integration Solutions with OGSA-DAI Page 109 3 GridNet2 Reports 3.23 Mr William Lee - Imperial This workshop presented the new features and capabilities of OGSA-DAI and provided an opportunity for discussion between users and developers. Others Dr Krause also attended the keynote speech by Steve Tuecke. Progress ByteIO interoperability testing Interoperability testing for ByteIO implementations started on 20th July 2007. Four implementations of the ByteIO interfaces participated in the tests. An experiences document was created during an OGF21 session. It was expected to be concluded by OGF22. Each implementation provides comments and experiences to the document. 3.22.5 OGF22 General Sessions • Opening session • Keynote: Charlie Catlett: What OGF Can Do for Enterprises (A View from the CIO Office) • Keynote: Irving Wladawsky-Berger: Cloud computing, Grids and the upcoming Cambrian Explosion in IT • Keynote: Addison Snell: Grid Usage and Productivity in HPC • Jen Schopf: The Encyclopedia of Life: A Web Page for Every Species Working Group Sessions OGSA-DMI This session covered an overview of the specification. There were several implementations of the specification and an interoperability fiesta was planned to take place before OGF 23. OGSA Data Architecture Future Directions The OGSA data architecture group was closed. The original aim of the working group had been achieved and it was agreed that there was no effort available for further work and re-chartering of the group. DFDL There was a progress update and a description of the DFDL specification. SAGA DAIS WG Issues were presented by the OGSA-DAI development team currently implementing WS-DAIX and WS-DAIR specifications. An interesting discussion ensued and useful feedback was noted for the WS-DAIR/WS-DAIX recommendations. Dr Krause presented feedback on an implementation of WS-DAIX by the University of Columbia. The main issue was that their tooling did not support the WSDL document that is included in the specification. RDF-S Specifications Update There was an update on progress and discussion on the RDF-S specifications. This meeting was very useful. Workshops Data Management Workshop There were four sessions for the Data Management workshop. The aim of this workshop was to discuss issues and strategies in Grid data management from an OGF and SNIA perspective with a focus on file system and data movement. This workshop was useful and there were several presentations that were of great interest to Dr Krause. Meetings There was a SAGA+DAIS meeting with Andre Merzky and DAIS-WG members to prepare for the working group session, which then discussed the development of SAGA APIs for DAIS. Progress ByteIO-WG The participants of the interoperability experiment summarised their experiences. Dr Krause prepared a report from the OGSA-DAI/EPCC perspective. The contributions were collected in an experiences document which was awaiting comments by the contributors. It was to be submitted as an OGF document when all participants had reviewed it. 3.23 Mr William Lee Imperial The reports for William Lee are consolidated with Dr Stephen McGough’s reports. For more information, see section 3.25, Dr Stephen McGough – Imperial. For a summary of Mr Lee’s GridNet2 activities, see section 2.23, Mr William Lee – Imperial. The SAGA and DAIS session included an overview of the DAIS specification for the benefit of the SAGA members. 3.24 A SAGA API for the DAIS specifications was being planned. SAGA was use-case driven. They were not interested in mirroring DAIS interfaces on the clientside. DAIS use cases would be used to drive the SAGA API specification. Mark Leese’s reports are consolidated into Dr Steve Fisher’s reports. For more information, see section 3.12, Dr Steve Fisher – STFC. Page 110 Mr Mark Leese – STFC For a summary of Mr Leese’s GridNet2 activities, see section 2.24, Mr Mark Leese – STFC. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial 3.25 Dr Stephen McGough – Imperial Other Groups Other groups which were of interest included the following: Dr Stephen McGough is Technical Co-ordinator for the London e-Science Centre, based in the Department of Computing at Imperial College. • Dr Stephen McGough’s GridNet2 award and this report also cover the GridNet2 activities for Mr William Lee and Mr Vesselin Novov. GRAAP (Grid Resource Allocation Agreement) WG who were evaluating the use of JSDL as part of their WS-Agreement specification • OGSA EMS (Execution Management Service) team who were determining how JSDL fits into the OGSA architecture • SAGA group who wished to know more about the API that JSDL was exposing. For summaries of the GridNet2 activities covered by this award, see the following sections: • Section 2.23, Mr William Lee – Imperial • Section 2.25, Dr Stephen McGough – Imperial • Section 2.28, Mr Vesselin Novov – Imperial 3.25.1 GGF14 Dr McGough attended GGF14. His main areas of focus were as follows: 3.25.2 GGF15 & OGSA F2F Dr McGough and Mr Lee attended GGF15. There was also an OGSA F2F meeting at GGF15. Their main areas of focus were: • GRAAP • JSDL • GRAAP • OGSA • EMS • EMS • SAGA • CDDLM • JSDL • OGSA RSS • OGSA BES • BES (EMS, BES) JSDL JSDL The main reason for attending GGF14 was the activity of disseminating JSDL. The JSDL 1.0 specification was almost complete and was in public comment. Achieving this required a campaign of information to the rest of the GGF community to both read and comment on the specification, as well as look towards adopting it for their work. The specification was the culmination of many months of effort from a large number of contributors, significant numbers of which were from the UK e-Science community. The JSDL specification had passed its public comments period and the group was focused on dealing with these public comments. This had produced a number of clarification points for the specification and a number of comments about desired features which were not currently present. The first session focused on dealing with the comments and figuring out which could be corrected in the document and which were features for future versions. There was a strong presence in this meeting from EGEE who had a number of comments on the JSDL specification. These were addressed with some level of satisfaction. In the second session, there was a presentation by a new (fifth) implementer of JSDL from Intel. This showed that there was a major uptake of JSDL within the wider (industrial) community. Dr McGough had co-chaired this group since its concept at GGF5 and Mr Lee has been an active participant of the effort. Sessions were organised to present the JSDL specification and those groups which were early adopters of the JSDL specification. This gave Dr McGough and Mr Lee the opportunity to showcase GridSAM, the OMII-funded, Grid Job submission and monitoring service which they have been developing at Imperial College. Three other early adopters also gave presentations, along with two other working groups who identified how they could use JSDL in their work. This showed significant early adoption of JSDL. Discussion of future directions for the JSDL group also took place. OGSA-BES The OGSA BES (Basic Execution Service) has been developing a service to wrapper the JSDL document language. This was an area Dr McGough and Mr Lee were keenly supporting as this would increase the uptake of JSDL and also matched with their work on GridSAM. Dr McGough and Mr Lee were active in the OGSA-BES work. GRAAP The GRAAP group were continuing to integrate the JSDL specification within examples in their documents. RSS was continuing to define how to select resources within the Grid. This, again, tied in closely with how JSDL was defined and how GridSAM could be used. The EMS group focused on how to model the submission of a job to a BES container, either directly or through the use of a job manager service. These presentations showed a strong take-up within the GGF of both JSDL and BES. Dr McGough and Mr Lee were extremely active in the JSDL and BES groups. This was giving them the opportunity to influence the outcome of this work based on the experiences gained within the UK from GridSAM. Page 111 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial CDDLM SAGA CDDLM was defining a mechanism for deployment and provisioning on the Grid. The CDDLM group was keen to work closely with JSDL, using it both above and below their approach. Work continued with the SAGA group looking at the API level for job submission and in the RSS group for how to select which resources should be used. Dr McGough and Mr Lee also made a presentation on GridSAM at the OMII-UK reach-out session. During the OGSA F2F meeting, the work of EMS was continued as the group looked into how CDDLM could be integrated into the models proposed during the main meeting, and developed an overall architecture for the execution of Jobs in a Grid, bringing JSDL, BES and CDDLM together. This was an area where Dr McGough and Mr Lee had a lot to offer the community. 3.25.3 GGF16 & OGSA F2F The GGF16 meeting also included an OGSA F2F meeting. Dr McGough and Mr Lee attended GGF16. Their main areas of focus were: OGSA In the OGSA sessions and F2F meetings, Dr McGough and Mr Lee were able to take an active role in the information model sessions for BES and the OGSA Container. This work developed from their efforts with GridSAM. JSDL was also on the panel for the OGSA session, again showing that JSDL and the work Dr McGough and Mr Lee have been funded to carry out were gaining wide acceptance within the Grid community. OGSA-BES • OGSA • Grid Interoperation • SAGA • OMII-UK presentation • Grid Scheduling The BES group was rapidly moving towards a standard. They seek (as part of the interoperation work) to be able to provide a standards approach to job submission. This effort also fitted in well with the work of the JSDL group and that of GridSAM. Dr McGough and Mr Lee were well placed to contribute to the BES standard from their prior experience in this area and the ideas they could bring through from the UK community. • OGSA Standards Interoperation • RSS • JSDL • BES There was much interest at this GGF about standards interoperation, specifically in the area of how Grid infrastructure could be made to interoperate. There were two different groups looking at two distinct cases: JSDL The JSDL 1.0 specification document had been published (GFD.56), and the priority was to disseminate it to the wider Grid community. At this GGF, there were two sessions regarding JSDL: • The first session was a workshop to disseminate JSDL. The session included a tutorial on JSDL and the presentation of five projects which were adopting JSDL. Again GridSAM was presented as one of the early adopters of JSDL. This brings the number of projects which have presented their use of JSDL to seven, along with four other projects informing us that they are evaluating/using JSDL. A total of six GGF groups are now using JSDL as part of their work and there are eleven different Distributed Resource Management services which JSDL can be translated into. • The second session was a discussion on how JSDL should progress in the future. The material from the last OGSA F2F breakout session was presented and a discussion held. The consensus was that a parallel job application extension, software requirements and reservations/co-allocations were the areas where most people wanted JSDL to work on next. Work on a proposed parallel extension was presented during the session. Page 112 1. Using existing Grid tools (mainly Globus) and getting Grids working together 2. Using emerging standards (JSDL, BES) and using these for interoperation. Dr McGough and Mr Lee’s efforts were placed firmly in the latter category with work on GridSAM fitting nicely into the interoperation work. This again highlighted the huge take-up of JSDL and BES. Grid Scheduling Grid Scheduling was more removed from the efforts that have been completed with JSDL and GridSAM; however, these did relate to the efforts carried out as part of the ICENI project which Dr McGough and Mr Lee were involved with before. This had allowed them to feed the results from this effort into this research group. 3.25.4 GGF17 Dr McGough and Mr Lee attended GGF17. Their main areas of focus were as follows: • HPC Profile BoF • SAGA • OGSA • OGSA RSS • OGSA EMS 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial • JSDL • Shibboleth • OGSA Information model • GIN JSDL At GGF17, the main focus for the JSDL group was on the definiti1on of a parallel job extension. Five projects, including GridSAM, presented their extensions to the JSDL specification for parallel jobs. This information was considered and evaluated in terms of what the wider community would really require from parallel job submissions. This was work that McGough and Mr Lee were able to contribute fully to and provide insight from their prior work in developing a parallel extension for GridSAM based on requirements from their users. This also brought up a discussion on how JSDL documents should be extended as this would be our first extension. By the end of the meeting, the group had created a draft schema for a SPMD extension. HPC Profile BoF Mr Novov has taken over Mr Lee’s role at the London e-Science Centre as developer for the GridSAM project, along with Mr Lee’s work on the OMII security work. Mr Novov attended the GGF meetings in place of Mr Lee, particularly the security sessions at GGF in alignment with his work on security. Because this was the first meeting since the merger with the EGA, many of the sessions at this event contained a more significant amount of dissemination. JSDL The JSDL sessions focused on the dependant job proposed extension and the parameter sweep job. Both of these items have been requested by the UK e-Science community. Much progress was made on these items, although no consensus was reached and not enough detail emerged to make a draft schema. From work that happened at the last OGSA F2F meeting, focusing on the HPC use case, a BoF meeting was held to form a new group to work on interoperation between different Grids. This work would be based on profiling down the BES and JSDL specifications. This was highly relevant work for what Dr McGough and Mr Lee had been doing already and something that they were taking a full and active role within. The plan was set out to hold an interoperation session at SuperComputing 06 in November. OGSA OGSA-BES HPCP The next iteration of the BES specification was worked upon which removed some of the complexities that had been added from the previous iteration, thus bringing it back more to what was already the case with GridSAM. The information modelling group continued its efforts in modelling the BES service and an overall information model for the Grid. Again, this was relevant as a better way for describing resource requirements within a JSDL document. The work on HPCP basic profile involved mostly dissemination at this event as Supercomputing was only one month away and all groups were working hard on implementation. Dr McGough and Mr Novov took the opportunity to meet up with a number of the other HPCP projects and discuss their work. This was very useful in terms of development for the interoperation event. BES was also involved in the interoperation event and would use this to help finalise the document. Other Groups Other Groups The EMS group looked at the overall architecture and specifically how CDDLM would fit within this. Dr McGough and Mr Lee continued to feed into the efforts of the SAGA and RSS groups. Dr McGough and Mr Novov continued their liaisons with the SAGA and OGSA EMS groups, giving feedback from their work with GridSAM. 3.25.5 GGF18 Dr McGough and Mr Novov attended GGF18. Their main areas of focus were as follows: • HPCPBP • Information Model • SAGA • JSDL • OGSA BES • OGSA EMS • GGF Authz WG During the OGSA sessions, the material on ClassAds style resource matching was discussed with a wider audience. This seemed to reach general support. These ideas were discussed and extended in both the OGSA and JSDL sessions. This work is highly relevant for the future of GridSAM and Dr McGough and Mr Novov were able to make significant contributions to the development. OGSA Authz The Shibboleth project is developing an open source implementation to support inter-institutional sharing of resources subject to access controls. Since the targeted resources are Web, not Grid, resources, there had been a number of activities attempting to integrate Shibboleth in Grid settings. One such effort was a prototype system allowing NGS users to access NGS facilities securely through the Shibboleth mechanism. The workshops provided updates on this and other similar activities and followed on experiences learned from the integration efforts. Those outcomes were of interest due to Shibboleth being one of the access control systems considered for integration in Dr McGough and Mr Novov’s OMII security project. The feedback provided at the workshops was helpful in assessing the cost and feasibility of re-designing the OMII security project to use Shibboleth or continuing with Page 113 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial the initial, simpler, prototype based on the OGSAAuthz architecture using PERMIS, another access control system. The goal of Dr McGough and Mr Novov’s security project was to build an authorisation system for web services hosted in an OMII Container. As such the system architecture, security standards and protocolrelated extensions provided by the Authz-WG were of particular relevance to their OMII-Authz project. The project's architecture closely follows the design of the OGSA-Authz architecture proposed by the AuthzWG. 3.25.6 OGF19 Dr McGough attended OGF19, and his main areas of focus were: • GIN • OGSA HPCP • OGSA Security • OGSA EMS • OGSA Information & data modelling • OGSA Workflow • OMII Software Forum • Standards All Hands • QoS BoF OMII-Authz also uses SAML. • JSDL • Subsequent evolution of the SAML standard to version 2.0 GIN • Various ideas on integrating Shibboleth components in Grid authorisation systems • Possible adoption of another OASIS standard – the eXtensible Access Control Markup Language (XACML). The discussions in the session focused on the following: • Refining the functionality expected from different sub-components of the Authz architecture • Authorisation request-related extensions to the Security Assertion Markup Language (SAML) By adhering to the requirements of these standards and relevant extensions, the development efforts behind the OMII security work ensured greater compatibility between OMII-Authz and other systems based on GT4 and Shib-Grid. GIN As a job submission service, GridSAM provided facilities for transfer of data files through various protocols. GridFTP was one of the currently supported protocols with plans to include SRB in the near future. For this reason, the activities in the Grid Interoperation Now (GIN) Community Group were of significant interest to the GridSAM team. Among other activities, this group's focus was on the implementation interoperation in data management and movement (GIN-DATA) and job description and submission (GIN-JOBS). The GGF sessions were used to disseminate results of the group's interoperability tests. Of particular note were the GIN-DATA tests including three different technologies, two of which have direct relevance to the GridSAM development and maintenance and user support efforts: • The GridFTP technology that can be seen as lowest common denominator for data movement in Grids • The SRM as an asynchronous standard interface to Grid storage systems to provide virtualisation of storage resources • The SRB as a shared collection management system of multiple (distributed) storage systems and their properties Page 114 In the Grid Interoperation/Interoperability Now (GIN) session, not only were JSDL, BES and HPCP given top billing for their interoperability, but GridSAM was also given significant mention. This showed not only that the standards work with which Dr McGough’s team have been engaged was having a wide impact on the community, but also that GridSAM was seen internationally as a prime example of using these emerging standards. Feedback from the interoperation which happened at SC06 was presented; again GridSAM was seen as one of the production Grids supporting the new emerging standards. The session went on to discuss where interoperability would go in the future. This covered such area as security – where there was mixed enthusiasm for the HPCP current use of username and passwords – with many calling for more secure techniques to be used such as X509. As GridSAM already supported this, Dr McGough was positive towards this advancement. HPC Profile The OGSA HPC Profile session focused on the changes since SC06. This included new versions of the HPC and BES specifications in light of what had happened at SC06. The most significant change to the HPC profile was that of adding a security section to the profile allowing for the use of X.509 certificates. This was useful as this is the default security method used for GridSAM. There was also discussion of how to delegate credentials with one option being the use of a MyProxy service. The MyProxy service was already being used within GridSAM, and hence Dr McGough was able to contribute his experiences to this discussion. OGSA Security In the OGSA Security session, there was some discussion over the diverging security models between OGSA Security and the HPC profile security. The HPC profile group had been adopting a stopgap security model – user name and password – though OGSA Security is proposing a more complete security model. Since the HPC profile group was now addressing the more general security scenario, it was hoped this would be resolved. This was beneficial for 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial GridSAM as the GridSAM security model was more complete and the GridSAM team preferred not to support username/password. OGSA EMS The OGSA EMS scenarios document version 1.0 had been submitted now to the editor for review. This contained basic execution patterns which use both JSDL and BES, and the main models in this document were on how to submit a job to a resource. The discussion centred on how the group should progress for the next document. This included such things as information services, Usage Records and Resource Usage service. OGSA Information and Data Modelling Work in the OGSA Information and data model session focused on the use of JSDL and XQuery to perform job and resource matching. The proposal was to add a more complete information model which could be used within a JSDL2 document and consumed by a BES2 service. The idea, as discussed previously though extended here, was to add Capabilities and Requirements to both the JSDL document and the resource description document. These could refer to each other through the use of XQuery statements. This would thus allow for the BES2 service (or potentially some higher service) to match jobs and resources. This again shows the wide uptake of both JSDL and BES within the wider Grid community. OGSA Workflow For the OGSA Workflow session there was much discussion on how OGSA should deal with workflow. There was a proposal to do a simple workflow language – much in the same way that JSDL was a success due to its simplicity – where each “node” would be a workflow document or a JSDL document. This was felt to be a rat-hole that would consume large amounts of time without giving significant return. Three alternative approaches were proposed: 1. Investigate how to layer workflow on top of OGSA 2. Change JSDL, BES, RSS to support simple job dependencies 3. Do workflow interoperation There was significant interest in approaches 1 and 2 though none in approach 3 at present. Dr McGough was selected to lead approach 1 and Michel Drecher to lead approach 2. The idea of 1 was more in fitting with the overall way GridSAM worked within the wider ICENI II infrastructure and hence was more appropriate for this work at OGF. OMII Software Forum (QoS) in the Grid, he led a BoF session at OGF19 meeting on QoS. This allowed Dr McGough to present ideas on QoS to the wider Grid community to see if there was enough synergy to take this work forward. There seemed to be a lot of confusion within the wider (outside of OGF) community for the meaning of QoS. Within networking, QoS has a very specific meaning and, as such, members from this community were unhappy with the use. Although there was quite some interest within the community on QoS issues, there was little in common between the two areas. It was decided that this should stay as a design team within OGSA until such time that there would be more people working on the same issues. JSDL Although the JSDL sessions at this OGF were not until the final half day of the meeting, over 30 attendees came to both the sessions, which is a large number for a session at any time. The focus of these sessions was on the following: • JSDL errata document • JSDL 2.0 discussion • Reaction to the HPC Profile draft document • Parallel application extension • Proposal for a parameter sweep extension The errata document was feeding back corrections into the JSDL 1.0 specification which have come in since the document was published, mostly from people who had implemented it. The intention was to keep compatibility with JSDL 1.0 by not providing any changes to the schema document, but instead providing extensions where needed and clarifying the text within the schema document. The JSDL 2.0 discussion focused on the implications of the work presented in the OGSA Information model session. A number of tracker items from the SPMD (parallel jobs) document were discussed and resolved within the session. A presentation of the HPC profile was made and discussed within the group. It was decided to roll a number of the clarifications that the HPC profile group had made into the JSDL errata document. There were a small number of changes to the schema which were left out as these would break interoperation with JSDL 1.0. A presentation was made for a proposal to do a JSDL parameter sweep extension. This was of keen interest for the Dr McGough as users of GridSAM have been asking for such functionality. The idea would be to use XPath statements to mark parts of the document to be changed along with lists of what that part of the document should be changed to. 3.25.7 OGF20 Dr McGough and Mr Novov attended OGF20. The main areas of interest for them were as follows: During the OMII Software forum, GridSAM was presented to the wider Grid community. Although Dr McGough’s team were not presenting this work themselves, they were able to attend the session to answer questions and discuss GridSAM with members of the audience. • Joint session on information modelling and computing resources • JSDL • HPC Profile QoS BoF • OGSA Workflow Because of the work in which Dr McGough has been active for the GRIDCC project on Quality of Service • Workflow sharing Page 115 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial • Grid Scheduling Architecture HPC Profile • OGSA BES • OGSA RUS • OGSA EMS • GridNet2 meeting The HPC Profile group was now working hard on producing the implementation experiences document. This was an area that Dr McGough and Mr Novov were active within, both as developers of JSDL and BES and also as active members of the interoperation activity at SC06. • OGSA RSS • Service Level Terms • Authz • Grid in the Distributed Computing Landscape • OGSA F2F Information Modelling and Computing Resources The joint session on information modelling and computing resources was held to bring together those people from the OGSA information modelling/resource management design team, the GLUE group and the Reference Model group. As these groups were now all working on similar areas, it was essential to prevent duplication of work. The groups were asked to describe what their work was and how it would fit together with the other emerging standards. In all three cases, JSDL and BES were identified as things that they would wish to interact with. This shows that JSDL and BES now have major adoption within the Grid community. Most groups were now feeding their requirements into the JSDL work and would help in the development of JSDL 2.0. As this work was highly relevant for McGough and Novov’s development of GridSAM, they were active within this session and were following up on this effort. JSDL There were two documents in public comment that were promoted at the JSDL sessions: • HPC profile application draft • Parallel application (SPMD) extension There was, therefore, a reach-out effort at OGF20 to encourage people to read these documents. For the JSDL 1.0 errata, there was feedback from a new project which had implemented JSDL (KnowARC), bringing the number of known implementations to eleven. This led to some new tracker items and some of the existing tracker items being resolved. In most cases, this was by tightening the definitions within the document based on feedback. During the second session, the ideas behind using XQuery were presented to the group, along with proof of concept examples showing that they could be used in this context. The parameter sweep extension was discussed further. In our final session, the group discussed the JSDL 2.0 extension. During the opening session of OGF20, the JSDL group was awarded with the Open Grid Forum Leadership Award. This award was given out to individuals or groups within the OGF for outstanding leadership and many contributions to the OGF mission. This was a significant acknowledgement from the community of the significance and the value of the work the JSDL group had achieved. Page 116 OGSA Workflow In the OGSA Workflow session, the two approaches (use existing workflow tools or design own workflow language) were further discussed. Some scoping activity was carried out on the group. Error handling was placed out of scope while error detection was in scope. Graphical representations and cyclic declarations were also placed out of scope. The idea of using BPEL over JSDL was proposed and the OMII-BPEL project, which uses GridSAM as an endpoint for BPEL activities, was presented. It was asked “if BPEL is just endorsed, who will complain?”, though this did not receive unanimous support. It was decided that a survey of existing activity was required before this work could continue further. This work fitted in nicely with Dr McGough and Mr Novov’s BES/JSDL work, but also with the general workflow and components work they were doing as part of the ICENI II development. Workflow Sharing By contrast, the Workflow Sharing session was more concerned with how workflows could be shared between Grid infrastructures. This could be through workflow translation or through the ability of one workflow system to execute a workflow in a different architecture. This was again relevant due to Dr McGough and Mr Novov’s work on the ICENI II development. Grid Scheduling Architecture The GSA-RG was working on interoperability of Grid scheduling entities. This was significant for the work with JSDL and BES because a scheduling entity would be the service built above a BES instance. Dr McGough and Mr Novov were able to contribute their knowledge from GridSAM, BES and JSDL into this work. OGSA BES For OGSA BES, the effort was now towards writing an experience document, to which Dr McGough and Mr Novov were contributing. OGSA RUS The OGSA RUS-WG had been inactive for some time as it had already published a specification. However, in light of changes within the community, the development of new standards and implementation experience had led to the group being revived for update. Through prior work (carried out by William Lee), Dr McGough and Mr Novov had feedback to give and had been active in this area. Much of discussions in the OGSA-RUS-WG session emphasised some fundamental scalability problems occurring with very large data sets of usage records and their transfer between clients and services and how the RUS and UR specifications could be improved in that respect. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial Dr McGough and Mr Novov shared similar experiences they had through the development of GridRUS. The GridRUS project was funded by the OMII-UK Managed Program. It was effectively a prototype implementation of the RUS draft specification providing query-able persistent storage of Usage Records. Like the services developed by some of the other participating groups, GridRUS is a WS-I-compliant web service, therefore prone to the discussed workload problems. However, much of the advantage of GridRUS rested in its modular architecture which allowed different components to be configured and used separately, if needed, by the user. The project suite consisted of the following: • The core engine providing a Java API for managing persistence and query of usage records job launching and file staging plug-ins • A web service implementing the Resource Usage Service specification • A package of client APIs and command-line tools for inserting or querying usage records stored in a RU Service • A package of XML to Java bound classes representing the RU Service and Usage Record schemas Should OMII-UK decide to fund further any development effort in this project, GridRUS would be in a very good position to become one of the first fully compliant RUS implementations and a benchmark product, given its highly flexible and reusable architecture. OGSA EMS The OGSA EMS group was focusing on how to discover information about the Grid for such things as job submission. In their model, jobs are described in JSDL and submitted to a BES instance. The work of the group is now to define how information can be obtained to achieve this. This was highly relevant work as it fits in with GridSAM and the work on BES and JSDL. GridNet2 Meeting Because this event was UK-based and, as such, most GridNet2-funded people would be attending, there was a GridNet2 meeting held during OGF20. This was to check up on the status of GridNet2 spending and to determine how to continue the effort of GridNet2 and promote its achievements. During this meeting, Omer Rana and Dr McGough were tasked with organising a GridNet2 workshop at the next OGF meeting. OGSA RSS The OGSA RSS group was working on the problem of how to select the resources to use within the Grid. This relied heavily on the JSDL description which could be used by the candidate set generator to list those resources which could execute the JSDL, based on resource requirements. This was, again, another scenario where JSDL was being used within the Grid architecture. terms that would be needed when discussing job submissions and file transfers (BES and ByteIO). This was significant work because it was something that could be made use of in future versions of GridSAM. OGSA F2F At the end of the main OGF event, a one-day OGSA F2F was held. The relevant items in this meeting are provided below. The OGSA EMS Scenarios document had been completed. The document discussed how to submit jobs to the Grid and relied on JSDL and BES. The workflow discussion from earlier in the week was continued. It was pointed out that other standards bodies did exist and other workflow standards did exist along with BPEL, such as XPDL although XPDL was more abstract in nature. Examples of other systems were given including Microsoft WF. In conclusion, it was felt that a primer/best practices document was most appropriate. Dr McGough and Mr Novov would contribute to this. OGSA Authz WG Over the period of two Authz-WG sessions, the group reviewed the status of some previously proposed and some new profile specifications. More text was added to the XACML Profile to cover obligations. The text explained exactly how gridmap files could be replaced with obligations, and how co-ordination decision making could be enabled with obligations. The model and diagrams of the WS Trust Profile were updated and made more generic. Protocol flows were still left the same but more possibilities were allowed. Some of the effort was allocated to the definition of the SAML V2.0 Profile for Virtual Organization attributes. Attribute names to be included were Virtual Organization, Group and Role, essentially replicating the attributes from the XACML Attribute Profile. Of much interest, related to Dr McGough’s OMII Authorisation project, were the reviews of a couple of authorisation services of which the implementations follow the XACML specification. G-Pbox, an authorisation service based in gLite, had taken an approach similar to the one in the OMII Authz application; one of its main component is an XACML-compliant PDP and the service interface processes xacml-context:Request against a repository of XACML policies and returns and xacml-context:Response. The shared similarities with other projects gave Dr McGough the confidence that the OMII Authz service project is evolving in a direction which might, in the near future, allow Dr McGough and Mr Novov to take part in a wider effort of testing/demonstrating interoperable authorisation services between EGEE, OSG, Globus and OMII. The GridShib Project continued its focus on a hybrid security token called X.509-bound SAML Token, that is, a SAML assertion bound to an X.509 certificate. It would be either a short-lived end-entity certificate or a proxy certificate. The resulting X.509-bound SAML Token Profile would be a straightforward extension of the WS-Security X.509 Token Profile, and therefore an implementation of the latter would be automatically an implementation of the former. Service Level Terms BoF The Service Level Terms BoF was held as a follow up to the QoS BoF held at the previous OGF meeting. The intention of this BoF was to scope in depth those Page 117 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial Grid in the Distributed Computing Landscape This was a generally informative session on current state of Grid. The session explored the way Grid infrastructure solutions had been used by organisations around the world to enable the modern knowledge-based, global economy. The discussion reviewed the role of Grid technologies within the broader realm of distributed computing and defined the common categories of Grid in use today. The presentation traced through the milestones of the evolution of Grid technologies from applicationspecific solutions to dynamic, shared and serviceoriented infrastructures. The session's main focus was a number of major trends in the use of Grids today, as follows: • The use of the infrastructure to reduce cost of operation, increase efficiency, optimisation and interoperability. • The use of the infrastructure to manage workload and services across multiple business units within an organisation. • The use of the infrastructure to enable interorganisational, service-oriented collaboration, within the boundaries of virtual organisations, through the adoption of common standards and practices. 3.25.8 OGF21 Dr McGough and Mr Novov attended OGF21. Their main areas of focus were as follows: • RUS • GSA-RG • JSDL • Joint Session on Information Modelling for Computing Resources • RSS • GridNet2 Workshop • OGSA Workflow • OGSA F2F, HPC Profile • SAGA • OGF tutorial on Vulnerability Assessment • Secure Coding Practices for Middleware RUS-WG The RUS-WG, as outlined previously, was re-working its published specification, based on implementation experiences. These included dealing with large records and the ability to request sub-parts of the whole Usage Record sets, along with the rendering of RUS both in WS-I and WS-RF. As Dr McGough and Mr Novov had previously implemented the old specification for RUS, they were keen to give feedback on their experiences on this work. GSA-RG The GSA-RG was focusing on scheduler interoperation. In their architecture, they saw BES Page 118 instances as being the end points for execution and JSDL as the language to describe the jobs sent to these resources. For this work, they sought to profile JSDL in terms of which attributes within the document are relevant for the scheduler and how they should be interpreted. This again showed further groups within the wider Grid community adopting JSDL and BES. The work of GSA was not to use the entirety of JSDL but to partner it with a scheduling language for those things relevant to scheduling but not to JSDL. McGough and Novov were able to take an active role within this session, both as a team that was developing scheduling above BES/JSDL through GridSAM and also as part of the JSDL group with knowledge of what was currently in JSDL and what would likely be there in the future. JSDL-WG There were three sessions for JSDL at this OGF. The first session was a general session for outlining the JSDL-WG work and current status. The group went over the JSDL 1.0 errata document for which they had resolved 25 of the 32 issues. Extra issues had been raised since the last F2F meeting. The JSDL group was also collecting experience reports from those groups that have implemented JSDL. At the time, they had four responses, with more promised. There was also a presentation from GridWay showing their use of JSDL, which meant there were now twelve known implementations of JSDL. The second session was focused on the XQuery extension for JSDL. This focused on the new idea to use capabilities and requirements in a JSDL document and how you could define the requirements as XQuery requests over the (to be matched) resource description document. This session looked into how XQuery could be used to achieve this goal and whether XPath, as a simpler mechanism, could be used to achieve the same result. The conclusion was that if all that the group wished to achieve was to extract parts of the other document, then XPath would be sufficient. However, if they wanted to perform more complex operations on the other document, for example, where the processor speed is twice as important as the memory, then the full XQuery was needed. The third session was more like a BoF session. The group was proposing to start some new work on an Activity schema. This document, for which JSDL would be a part, would contain the whole activity lifecycle of the job. Information on resource usage, scheduling, job status would all be placed into this document, although the contents of these sections would not be defined by this specification, merely refer to the appropriate specification. There were presentations on GridSAM, UDAP, GLUE and NEREGI as to what each project has already covered in this area. This session allowed Dr McGough and Mr Novov to give a presentation on GridSAM and place it in the limelight within the community. The Joint Session on Information Modelling for Computing Resources brought together most of the parties that had discussed information models at the last OGSA F2F meeting in a wider context. The discussion was about how GLUE relates to other specifications such as BES and JSDL. This allowed the groups to fill in the details of how they were all integrating their work together. For the JSDL group, the focus was adding resource descriptions in the JSDL document from those in GLUE. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial The Resource Selection Service (RSS) is now submitted to the OGF editor. This specification defined an interface for how to select a set of potential resources for executing a job. As such, it relied heavily on JSDL as a job description language and for terms defining the type of resources that are appropriate. This again showed uptake of JSDL (and BES) throughout the OGF. GridNet2 Workshop For this OGF, Dr McGough had organised a GridNet2 Workshop in collaboration with Omer Rana. This allowed all participants funded through GridNet2 to highlight the work they have achieved. Dr McGough and Mr Novov were able to present work on GridSAM, GRIDCC and the development of the ICENI II architecture as areas which have both influenced the development of the OGF standards, and also benefited from these emerging standards. As much of the scope for the OGSA Workflow group had changed since the last OGF meeting, there was a presentation of an overview of its progression and metamorphosis into its new form. Out was the plan to develop a workflow language or to add simple workflow constructs onto JSDL. Instead, the group was focusing on surveying what was currently being used within the wider community. As such, the initial results of the survey were presented. This led to discussion on what else the group should be asking and how to solicit responses from other groups. It was also discovered at this stage that the WFM-RG is currently performing a similar survey and it was decided to combine the effort. OGSA F2F At the end of the OGF meeting a one day OGSA F2F meeting was held. An EMS session was held to discuss the revising of the EMS scenarios document. The plan was to swap out BES for HPCP, where appropriate, in the document and downplay CDDLM as it is now “resting”. Other work, such as RSS, needed to be reviewed. It was decided that, as there was not yet enough community support for the Service Level Terms work and as there was not enough consensus for what to do in this area, it was not a prime area for standardisation. As such, the work was left until the group had knowledge on what could be done and what people wanted to do. In the area of workflow, it was decided that rather than duplicating the work of the WFM-RG’s survey, OGSA would collaborate with them on their survey, and once completed, determine how best for OGSA to progress with workflow. The Data Movement Interface was an interesting piece of work which helps abstract the user away from how data was actually transported around the Grid. Rather than specifying protocols and resources, the user specifies known endpoints – one for where the data is currently and one for where you wish the data to end up. The DMI was then responsible for identifying the most appropriate method for transferring the data. This had major implications for both JSDL and GridSAM. DMI could be used to describe the data staging endpoints and GridSAM could use DMI to source and send the files. This was highly relevant work. HPCP-WG Much of the work in the HPC Profile WG session focused on the upcoming Interoperability fest scheduled to take place at the SuperComputing convention in November'07. In particular, much of the discussion covered the profile extensions which were to be demonstrated as new additions to the specification. The Data Staging extensions were based in large part on the existing Data Staging element in the JSDL standard, but the extension document restricted the set of options otherwise acceptable under the original specification. The text of the extension explicitly listed the type of data transfer protocols which compliant implementations would be expected to support - http, ftp (both supported by GridSAM) and scp. The Activity Credential element would provide a security context to the processing of a submitted computational job. The inclusion of this element was to account for the missing, but otherwise necessary, security left out of the scope of the JSDL standard. At the session, the group agreed on the general framework, sequence and logistical details of the interoperability demonstration which in the most part would replicate the successful interoperability fest at the SuperComputing in the previous year. The functionality provided by the comparatively more mature GridSAM already offered staging of input/output files, as well as a particular service configuration option, allowing the user to supply security credentials necessary for data transfers and/or the launching of the executables. SAGA-RG Simple API for Grid Applications (SAGA) is a C++ and Java implementation of the GWD-R.90 standard specification. The efforts of the SAGA group are focused on the development of a high level framework for use in creating Grid-aware applications. Dr McGough’s interest in the activities of this group came from the fact that this API is presented to the user as a simplified way to interact with Grid-middleware products, with one such being GridSAM. In designing and implementing the recommendations in GWD-R.90, the group addresses the inherited difficulties and issues of heterogeneity and the complexity of Grid applications. Therefore, SAGA was envisioned to provide common Grid functionality at the correct level of abstraction, along with the ability to hide varying underlying semantics. SAGA was able to make an important contribution to interoperability at the application level (an abstract layer above the one where middleware, such as GridSAM, functions), especially given that the SAGA software environment had matured sufficiently, thanks to advances at several levels: the SAGA engines (C++ and Java); the adaptors for different middleware distributions (Condor, GridSAM); and the maturity of the core SAGA specification. During the group session, the discussion revolved mainly around the current status of the two programme language bindings and the road-map deliverables, as well as the specific ways SAGA contributed to interoperability efforts among Grid users, application and middleware developers, and resource providers. Vulnerability Assessment and Secure Coding Practices for Middleware The OGF tutorial on Vulnerability Assessment and Secure Coding Practices for Middleware gave McGough and Novov valuable insight into the lowlevel requirements for a secure software development. Secure programming would be essential for their OMII Authorisation application. The tutorial was relevant to anyone attempting to assess software for security flaws and for developers Page 119 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial trying to minimise security flaws in software they write. Part of the tutorial covered a process for actively discovering vulnerabilities by: • OGSA Information • OGSA EMS 1. The gathering of information about a targeted system. • Cloud Computing • GSA • GRAAP • OGSA Authz 2. The use of the collected information to direct the search for vulnerabilities. 3. The integration of the above two steps into the development cycle. The tutorial pointed out specific coding practices for prevention of vulnerabilities, exposed many types of vulnerabilities with examples of how they would commonly arise, and showed some simple techniques for avoiding them. OMII-UK Workshop At the OMII-UK workshop, there was a presentation of the overall direction of the organisation's current activities. OMII-UK's mission had been to provide the UK e-Science community and their international collaborators with the software solutions and support they needed to enable a sustained future. During the session, there were demonstrations on some of the ways to help research communities benefit from the increasing number of resource providers - from national Grid infrastructures, such as the NGS, to campus Grids and departmental clusters. The demonstrations also included examples of using stable APIs (for example, SAGA) and showing how common scenarios can be solved by integrating components and providing new access mechanisms such as portals, virtual research environments, and desktop Grids. There was also discussion about how projects could build on top of efforts such as GIN and OMII-Europe to bring together Grids through open standards and interoperable implementations. The future of the OMII-UK and new developments in the organisation's efforts are of a particular importance to Dr McGough’s team in London e-Science Centre, who have had a very productive relationship with OMII-UK. A number of projects had already resulted from this relationship, and some of them had already proved successful in delivering Grid middleware applications to the wider scientific community, specifically: • GridSAM • GridRUS • OMII-Authz • GridBS 3.25.9 OGF22 Dr McGough and Mr Novov attended OGF22. Their main areas of focus were as follows: • OGSA-RUS • SAGA • JSDL • WFM-RG • AIDS • HPCBP Page 120 OGSA RUS-RG With new reinvigoration of effort within the OGSARUS group, there were now new documents coming out from this group. Implementations of this work are in progress and interactions with new standards are emerging. As Dr McGough’s team wrote the OMII-RUS implementation, this is a prime area for them to work within, and they are giving feedback on their experience to inform the OGSA-RUS work. SAGA-RG The SAGA group was defining an API for accessing the Grid. This was of interest to Dr McGough and Mr Novov as one of the first areas for which they were developing their API was job execution. This was not seen as a competitor to JSDL and BES, but rather a layer which could be built on top of these specifications. Coders could use the SAGA API to define jobs for execution, and then the implementation of SAGA uses JSDL and BES to execute the job. This was significant to Dr McGough and Mr Novov as GridSAM had an API for developing JSDL documents and submitting them which could be extended to include the SAGA API. JSDL-WG The JSDL group held two main sessions, along with a BoF session for the Application Instance Document Schema (see below). The JSDL 1.0 Errata document had completed the public comment phase and was now awaiting evaluation from GFSG as to final publication. This was significant as it was the culmination of over twelve groups’ evaluation through implementation of the schema, indicating major uptake within the community. During the general session, the JSDL group discussed the work that was done previously for the parameter sweep extension. This was proposed around one year ago, though efforts on HPCP profile and the errata document had caused this work to be suspended. Through interest in uptake, mainly from GridSAM in the UK community, this work was revived and presented again. The document was effectively complete and received positive feedback from those attending the session. The intention was to tidy up the document and submit it to GFSG. In the second session, the JSDL group further discussed the Application Instance Document Schema. For this, Dr McGough and Mr Novov were able to present more information on how they were using instance information within GridSAM. It was proposed that the scope of the work was too much, although it was clarified that this would just be where to put the data and not what the data was, not even to profile the work of other standards. The consensus was to continue this work within the JSDL group by scoping out exactly what was needed – use case gathering – and then evaluate how to take it further. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial WFM-RG Because of his role in the OGSA design team for workflow, Dr McGough attended the WFM-RG session. The WFM-RG was focusing on interoperability in workflow and conducting a survey of to find out what workflow systems and use cases existed already. It was noted that there was already a large amount of effort outside the OGF on workflow standardisation, and the intention here was to survey this before moving forward. This was of interest both from the point of view of OGSA workflow and from Dr McGough and Mr Novov’s work with ICENI II and GRIDCC. Application Instance Document Scheme BoF The Application Instance Document Schema BoF allowed Dr McGough and Mr Novov to present their instance extensions to the JSDL specification within GridSAM. The intention of this group would be to extend (probably outside) JSDL to encompass all the other information about a job which is not part of the jobs submission description. This would include information about scheduling, resource usage, and security credentials used. The information here would be dynamic and might not all be located in the same document but could just refer to other documents. It was not clear at the end of the BoF how the work should continue. However, these issues were resolved in the other JSDL sessions. HPCBP After the second successful HPC Profile interoperability fest at SuperComputing '07, the HPCBP group held a session on specification adoption. This session outlined the history of HPCBP, through the work of JSDL, BES and finally HPCBP. This clearly showed the significance of the work in which Dr McGough had been involved through efforts in JSDL and BES, and the uptake within the community, as the presentations highlighted both industrial implementations of these specifications, as well as those from research groups such as Dr McGough's team. The HPCP-WG continued improving the specifications of the extensions to the basic profile. Since the integrated HPCP support in GridSAM was one of the early implementations of the profile, Dr McGough and Mr Novov had participated regularly at the WG's weekly conference calls as well as all OGF working sessions. activities) expanded the basic filtering by allowing for scoping based on UserName, Owner, State etc. An HPCP implementation may not support the AFP, but must return UnsupportedFeatureFault if it doesn't. The proposed Advance Filter (for resources) expanded the basic filtering by returning available resources in per-resource-type records rather than per-resource-instance records. A particular issue was raised about the use of String-based jobIDs in the returned data of the AFP calls. The jobID was returned inside EPRs and the EPRs were opaque to the user. (The jobID value format, as well as its placement within the EPR, is service-specific). Application Template (AT) The purpose of the AT was to simplify the job submission with JSDL documents, so that the user would not need to specify every detail, executable, parameters etc, when submitting a job. However, there needed to be a way of discovering the ATs available at given HPCP service end point. The proposed way was to use an ApplicationTemplates element in GetFactoryAttribtesDocument return doc. containing multiple JSDLApplication elements providing the ApplicationName, ApplicationVersion and Description values. A point for further discussion was the fact that as proposed the AT’s names were not portable among HPCBP endpoints, to use an AT, a client would provide a jsdl:ApplicationName (optional jsdl:ApplicationVersion) element and could specify jsdl:JobIdentification, jsdl:DataStaging elements; jsdl:Resources elements that were over-ridden by the AT. The precedence of HPC application values after the HPCP endpoint applied an AT to a submitted JSDL is predefined. Some of the values would be replaced by values defined in the AT, but others would be added. Another potential problem that the group needed to address was the 'silent' replacement of JSDL values supplied by the user which clash with values defined in the ATs. A possible solution was the use of a set of faults used to notify the user what exactly was/was not accepted. File Staging Profile (FSP) • Advanced Filter Profile (AFP) • Application Template (AT) The FSP was one of the first extensions to the basic HPC profile and, given the fair amount of work that the WG has put into it, had reached a more mature stage. The profile was moved to public comment period from 4 -11 Feb '08 but this period was extended. The profile, as it stood, was based on the DataStaging Element in JSDL v1.0 with the addition of a 'Credential' element, specification of a narrow set of supported protocols, discovery mechanism through BESExtension element and file staging failure exceptions. • File Staging Profile (FSP) Kerberos Authentication The following three extensions underwent revisions: At the OGF22 group session, the discussion focused on the latest state of these documents and a review of the work completed in the months after SC07. Advance Filter Profile (AFP) The purpose of the AFP was to define a functionality for scoping the data returned by calls to GetFactoryAttributesDocument(). The current returned document definition allowed basic filtering by simple inclusion or exclusion of given activities and resources. The proposed Advance Filter (for Along with the review of the three HPC profile extensions, much of the session discussion focused on the need for support of Kerberos Authentication within the Credential element of the File Staging profile. This particular requirement was coming from users of vendors implementing the HPCP. There seemed to be a large user base using Kerberos as authN mechanism, single-sign-on services built on top of Kerberos, and/or a large number of batch system relying on Kerberos credentials. In addition to authN of services, the delegation of access rights was seen as an issue to be addressed. Kerberos authN Page 121 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial mechanism was simple and well understood; however, it is usually non-trivial to implement and program for. HPCBP Developments and GridSAM One of the latest developments within the WG would likely affect the current plans for the next GridSAM release. There was a proposal from the GenesisII team with respect to integrating the task of application deployment within a BES endpoint functionality. It had been the team's experience that the use of the jsdl:DataStaging element for staging executables and configuration files was inefficient and didn't scale. The team's evaluation of the CDDLM and some other similar deployment mechanisms found them to be too complex. The evaluation process found that most of the use cases involve a transferring and unpacking zipped binary executable and/or configuration files. The proposed solution would introduce a new deployment descriptor element of the JSDL document and let the BES endpoint handle this task. GridSAM had supported the staging of input/output files as part of a computational job processing since its inception. Should this new proposal be accepted and become officially a part of the JSDL standard, GridSAM was in a position to become one of the first BES endpoints to provide such application deployment functionality. OGSA Info The OGSA Information Model was close to submission and contained significant input from the JSDL group, both in terms of the new matching of resources and jobs paradigm and the development of XQuery interactions on the documents. OGSA EMS The OGSA EMS specification was further discussed. CDDLM was being removed from the specification, where appropriate, and file staging and resource selection was being extended. This fitted in nicely with the work Dr McGough and Mr Novov were doing with GridSAM and GridBS and they were participating in the OGSA EMS effort with their experiences from this work. Cloud Computing An open BoF session was held on Cloud Computing and how this fitted into the OGF model. There were a number of presentations about different Cloud implementations and how these were partly Grids and partly something at a higher level. The general consensus from the session was that Cloud computing was a higher level above the Grid – more the interface rather than the “how it works”. The feeling was that Cloud computing was relevant for OGF although, at present, this was not an area that OGF could standardise. The work that Dr McGough and Mr Novov had been doing on JSDL, BES and GridSAM, however, seemed quite in line with the Cloud model as they provided an interface for submission which could be used with the Cloud. GSA-RG The Grid Scheduling Architecture group was working towards interoperability in and between scheduling services on the Grid. They sought to take JSDL documents, along with scheduling documents, to select the appropriate BES instance to deploy a job to. If in the case the scheduler can’t find a suitable BES instance, it can communicate with other schedulers Page 122 to find a resource. Through Dr McGough and Mr Novov’s work on GridSAM and GridBS, they were able to contribute their experience of these scenarios back to the group. GRAAP-WG The GRAAP group had two implementations of services for submitting jobs through WS-Agreement which were able to interoperate. There was, however, much concern during their session that what they had implemented was far more complex than merely implementing an interoperation for WS-Agreement. Their implementations both relied on JSDL, among other specifications, and although this was good from the JSDL group’s point of view, as they were effectively providing another two implementations and showing interoperation of JSDL for the group, there was concern that this would not help others wishing to implement WS-Agreement for other use cases. OGSA Authz-WG Due to Dr McGough and Mr Novov’s ongoing development of the OMII Authorisation project, they had been closely following the Authz WG's and other Grid security-related activities in the last few OGFs. Dr McGough and Mr Novov’s design was based in large part on the proposed OGSA Authz architecture. The latest changes and the current state of this architecture were reviewed, as well as the recommendation of removing the OGSA acronym from the official specification title. The architecture and its document were deemed to be complete and ready to be moved to public comment phase. The proposed XACML profile was also reviewed. The profile would be stripped of some now-mandatory attributes passed between Policy Enforcement Point (PEP) and Policy Decision Point (PDP). The attributes were to be moved to the appendix and made optional. So far only one team had completed a reference implementation and published experience and test results. Support for XACML had been considered as part of Dr McGough and Mr Novov’s OMII Authorisation development effort and it was quite likely such support would be introduced in future releases. The group also discussed the SAML profile, paying particular attention to definitions for attributes to express VO-related data, such as cross-organisational groups and roles. However, an issue was raised that this particular profile/document type fell outside the charter of the Authz WG. Since there had been an administrative decision to close this WG, the document's future remained uncertain. One proposed solution was to submit a new charter document along with the proposed profile by interested parties. If, however, this approach was to be undertaken, there would have to be a careful consideration of the amount of work needed to bring the profile to a final state. On the issue of closing the WG, the timely submission of the Authz architecture, XACML and WS-Trust documents for public comment period should allow for enough time for any comment/suggestions to be addressed and any relevant changes incorporated into the final official versions. The tentative deadline for this should be the OGF in Barcelona where the last OGSA Authz WG meeting is scheduled. Any outstanding concerns raised afterwards were to be resolved by following the relevant OGF mechanism in practice. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial 3.25.10 OGSA F2F (12) • Different application types (java, SQL, parallel) Dr McGough attended the OGGSA F2F (12) meeting held on 15-19 August 2005, in Sunnyvale, California. His main areas of focus were as follows: • Job instance information • Quality of Service • OGSA-BES • Parametric jobs • OGSA-RSS • Variables • JSDL • Workflow • OGSA Resource Model • Job preferences After the first JSDL F2F meeting, it was discovered that there was a significant cross over between the JSDL attendees and the OGSA attendees. It was therefore determined that holding the JSDL working group F2F meetings as part of the OGSA F2F meetings was more appropriate. • Security During this F2F meeting, there was significant discussion of the RM (Resource Model) specification work. This was highly significant to the JSDL effort because the resource model currently described within JSDL is crude and basic, awaiting a more complete model. This allowed Dr McGough to help steer the effort of the RM team to those beneficial for JSDL, GridSAM and the UK e-Science community in general. Further discussion of the BES specification helped to iron out problems within the state model, thereby simplifying it. The OGSA RSS group are defining how to select resources to use within the Grid, which relates closely in with both the RM and JSDL efforts. Both of these areas were relevant from Dr McGough’s work on GridSAM. A breakout session was held for those interested in JSDL where a roadmap was developed as to how to progress JSDL post version 1.0. This session included extensions for parallel jobs, parametric jobs, job templates and other application types (for example, database queries). 3.25.11 OGSA F2F (13) Dr McGough attend the OGSA F2F (13) held on 16-20 January 2006 in Sunnyvale, CA. The main areas of focus were as follows: • BES • JSDL • EMS The JSDL 1.0 specification was published just before this F2F meeting. This placed JSDL as a recommended specification which was a significant accomplishment for the work of the group. Within the OGSA F2F meeting, there was a JSDL breakout session in which the group discussed the future of JSDL and in which areas the group wished to progress in the future. The group identified a number of key areas for which the group had received significant requests from the wider community, including members of the UK community. These issues included: • Data model • How to deal with extensions • Partial data staging The sessions also covered developing the BES specification and the EMS architecture, where discussion focused on which services talked to each other and which languages were used. 3.25.12 OGSA F2F (14) Dr McGough attended the OGSA F2F (14) held on 4-7 April 2006. His main areas of focus were as follows: • HPC Profile • EMS • CDDLM • Information Model At this F2F meeting, new effort was emerging in the area of interoperation. There was much interest in producing a High Performance Computing Profile of both the JSDL and BES work to allow large scale clusters to be exposed and interoperated between. From the previous work on JSDL, BES and GridSAM Dr McGough was ideally positioned to contribute fully in this area. The idea of this work was to condense these specifications into a profile that most projects could implement in time for Supercomputing 06 in November. As GridSAM had already implemented JSDL and BES, this made Dr McGough’s situation much simpler because OGSA already had full functionality. The organisations and projects involved in this effort included Microsoft, Platform and EGEE, which again showed the rapid take-up of the JSDL and BES standards. In the EMS sessions there was discussion of how to extend JSDL to work within the EMS architecture. The CDDLM session contained discussion of how to merge CDL (the language used to describe what needs to be deployed) and JSDL. The idea was to merge elements of JSDL and CDL to make a more complete deployment and execution language. Within the Information Modelling sessions there was discussion of how to map JSDL resource elements into the CIM model space and vice versa. This also led to a discussion of what needed to be extended within JSDL. 3.25.13 OGSA F2F (15) Dr McGough attended the OGSA F2F (15) held on 1720 July 2006, at Argonne National Labs in Illinois. His main areas of focus were as follows: Page 123 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial • Information Model • HPC Profile • BES • EMS • CDDLM At this F2F meeting, there was an emergent new theme about how data (information) should be modelled. This was the idea that for all services and consumers, each should not only have requirements of what they require, but also information about what they provide. This was very much akin to the approach taken within the Condor ClassAds language. Thus resources had capabilities and requirements of users, and users of a service had requirements and capabilities, such as identity of the user and what characteristics they possess. This fitted in well with JSDL and the plans for future development. But more importantly, this fitted in strongly with the work that Dr McGough had been doing within the OGF, as it was working on an XML version of Condor ClassAds which brought Dr McGough to the position of starting work on a job description language in the first place, having indepth knowledge of this area. The approach proposed would be to adopt more modern realisation of the language, using such tooling as XQuery and/or XPath, and rendering the document within XML. There was much discussion as to whether a strongly XML typed language would be preferable to a more weakly typed XML language where the strong typing of XML is not used. This new information model plan fed into the JSDL session in which it was decided to adopt a requirements and capabilities approach for the upcoming JSDL 2.0 work. This would allow for two-way matching within a (less basic) BES service and cater for the resource’s requirements as well as the users. This would require the use of either a more inelegant BES service or a broker service as an intermediary. From his work with GridSAM and prior work with the Condor matchmaker, Dr McGough was able to make significant input into this discussion. Other issues which were discussed for the process of moving from JSDL 1.0 to 2.0 included: • Moving items around the document to make it more about “what needs to be done” rather than “how to do it” • References between documents to reduce repeating common elements • The reification of JSDL documents from abstract to more concrete, which was something that users of GridSAM had at times desired The BES team had decided to postpone the Information Model of the BES container as this had become a blocking item for the progress of BES. Work would progress without this, using a very simple model instead, until such time as an information model became more mature. This was of some relief to Dr McGough as GridSAM already used a simple model which he had found sufficient for GridSAM’s users. Discussion also focused on the state model and how to deal with multiple invocations (nested or parallel). Page 124 The HPCP group presented a set of requirements over the BES and JSDL specifications which were blockers for the HPCP interoperation event. These included the removal of file staging for this iteration (as few projects could implement these in time), the removal of POSIX requirements on job submissions (so that Windows could be used) and the simplification of the BES state model. These issues were discussed and resolved. This again showed the significant impact of the JSDL and BES work on OGF and the importance of the group’s effort in this regard. The EMS architecture scenarios were discussed and improved upon and the CDDLM group discussed how to perform POSIX deployments with JSDL. 3.25.14 OGSA F2F (17) Dr McGough attended the OGSA F2F (17). The main areas of focus were as follows: • JSDL • Information Model • Interoperability • Security • QoS • Workflow JSDL At this OGSA F2F meeting, there was a JSDL session to investigate further ideas about JSDL 2.0, as discussed at the last OGF meeting. The ideas of pre and post conditions in a JSDL document were discussed along with the capabilities and requirements that a job would have to enable matching with resources. Concern was raised over pre and post conditions being a back-door route to allowing workflows. The aim of this session was also to determine how JSDL 2.0 would be specified, and its relationship to other groups in OGF. Dr McGough was able to contribute extensively to this session as it not only involves one of the groups that his team are most active within, but also allowed them to feed in requirements from the UK community. It was decided that the group should proceed on two tracks. The first was to develop extensions to JSDL 1.0 which can be used now, and the second was to draft up a JSDL 2.0 specification skeleton to allow development of some of these ideas. Information Model The Information Model had changed somewhat since the last session at OGF19. The GLUE schema had been brought into the OGF standards body and, as such, there was much interest in working with this. The GLUE and BES specifications had been taken as a starting point for work in developing the Information Model and this was presented as the starting point for the model. There was some concern over starting from GLUE although as it had been used in many Grid instances for some time now, this was as good a point to start from as any. As this related mostly to computational (BES) jobs, it was useful to take part in this session as Dr McGough was able to feed the requirements from GridSAM into this work. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial Interoperability 3.25.15 OGSA F2F (18) There was much discussion within both OGSA and OGF on the issues of interoperability. During this F2F meeting, the areas in which interoperability through standards could be achieved were discussed, and a list of projects which would be able to take part in interoperation tests was constructed. This enabled Dr McGough to propose GridSAM and other UK-based projects for these interoperation tests. Dr McGough and Mr Novov attended the OGSA F2F (18). Their main areas of focus were as follows: • Information Modelling • AuthN • EMS Scenario Security • JSDL One of the main areas which had still to achieve any significant standardisation within the Grid community was that of security. There were many groups working in this area although, at the time, there was little in the way of standards and interoperation. This was borne out by the recent HPC profile interoperations performed at SC06. The proposal was to define a simple security specification where complex issues such as delegation and authentication were not addressed, and in general, to leave out all the controversial issues which were preventing standardisation. This was good as far as Dr McGough’s work on GridSAM was concerned as GridSAM used a simple security model that was similar to the one proposed for this work. As the GridSAM security model was that of the main OMII container, this was useful for the work of OMII in general, and Dr McGough was able to help scope out what this work should be. • Workflow • Service Level Terms QoS In a follow-up to the QoS BoF held at the last OGF, Dr McGough was involved in running a QoS session at the OGSA F2F meeting. In terms of what the group was doing, it seemed as though SLA was more appropriate than QoS. The feeling was that if the group has continuous parameters, this would lead to too much choice at this stage and be too complex to deal with. The idea of the photo processing lab options (1 hour, 1 day, or 1 week) or the MacDonald’s menu approach would be easier to deal with at present. Though constraints (such as price) needed to be placed on these as otherwise everyone would ask for the “best” option all of the time. Although this did not fit directly in with what had been done in the GRIDCC project, it did have much more traction for standardisation, and as such, Dr McGough was able to contribute to this work. Workflow To continue the workflow efforts from the previous OGF, the group discussed the use of workflow languages such as BPEL. There was concern within the group about BPEL issues such as workflows being unchangeable once deployed, and workflows not being first class entities. Dr McGough was able to describe work that he had done through the GRIDCC project to alleviate the unchangeable nature of BPEL documents by changing endpoints dynamically. The general feeling was that OGSA should focus on ensuring that the architecture was easy to use for workflows and that the group could not, at this stage, rule out the need to write their own workflow language until they hadgathered information on existing workflow languages and the use of these. This was something Dr McGough was able to contribute to through his work on GRIDCC, and through using GridSAM as part of a workflow system. Information Modelling In the Information Modelling session, a discussion was held as to how the CIM, DM and GLUE models interact with each other; the JSDL specification and the BES model were also part of this discussion. The relationship between these entities was defined and discussion ensued as to how entities such as jobs and resources could be brought together. This led onto a discussion as to how JSDL 2.0 could be developed to have capabilities and requirements which would match with similar capabilities and requirements of the resource. It was decided that the requirements could be defined into two parts actual requirements and ranking/preferences. It was felt that, at the current stage, it would be best to limit this just to the actual requirements and that ranking of the possible matches could be done at a later stage. The idea of two-way matching (where both the consumer and provider have requirements which are adhered to) as opposed to one-way matching was discussed. Again the idea was to start with one-way matching as this would be simpler. The discussion moved onto how to achieve this with XQuery being proposed as a prime candidate. The consensus was that JSDL should work on its extension for matching using XQuery as a testcase for this approach. This was an interesting session from both a JSDL point of view, as Dr McGough and Mr Novov were significant players in the session, but also for GridSAM because this was an area where they would be developing GridSAM. AuthN The AuthN group had encountered the problem that most organisations and companies had already implemented their own security infrastructure, and as such, were unlikely to change this in the short term (if at all). This made the chance for easy interoperation slight. This was a major problem for Grid adoption, because without a common security model, interoperation at any other level causes difficulties. Work was ongoing to see if a common “simple” case can be defined. EMS Scenarios The EMS scenarios document had been published but it was outdated because it was based on old versions of some of the specifications and some specifications did not seem to be active any more. The original document focused on job submission (BES/JSDL) and the deployment of software using CDDLM. However, CDDLM had not been adopted within the community, because, in many cases, it was considered too complex to implement. As such, the intention was to revise the document to take account of HPCP and replace CDDLM with other standards where ones exist, or at least provide examples of Page 125 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial simple ways to do this, such as transferring a zip (tar) file to the endpoint and expanding it before executing the code. This also required the addition of file staging scenarios within the document. Other changes would include update of the Resource Selection Service text and revision of the Information Service in light of the recent changes. This was significant work for Dr McGough and Mr Novov, due to their work with BES and JSDL, and also as these scenarios match closely the scenarios provided in GridSAM already. This further showed the major uptake of their work with JSDL/BES within other standards groups. JSDL For the JSDL session, most of the discussion continued from the themes in the previous sessions. As the work for JSDL was becoming integral to the overall OGSA architecture, it was difficult to isolate this work to just its own session. The group had managed to resolve 24 out of the 29 open issues for the JSDL 1.0 errata document and it was hoped to be able to publish this document soon. For the requirements specification, there would be a close synergy with the work happening in the OGSA-RSS group. As such, effort from the JSDL group would go towards helping finish the OGSA-RSS specification. Work on the XQuery prototyping was progressing well and examples are being generated. Workflow At the previous OGF it was decided that the OGSA design team on workflow would conduct a survey of existing uses of workflow. In this session, the group discussed the possible scenarios for which they would ask questions and what the group intended to do with the answers once they had them. It was proposed that the group should focus on Web Services-based workflow systems as opposed to just any workflow systems. However, for JSDL many underlying systems used workflows which are not Web Services. The conclusion was to make a spreadsheet of known workflow systems and contact users and developers to fill out the details for the systems they know. Service Level Terms In the Service Level Terms session, the group followed up the BoF held at the previous OGF meeting. It was felt that the best approach would be to offer different tiers of support based on what the user was willing to “pay” for - such as Bronze, Silver and Gold. These would be pre-defined tiers that the user would select. This work did not seem ready for standardisation yet, though this information needed to be captured for future use. Again this was relevant for the work Dr McGough and Mr Novov were doing with GridSAM. 3.25.16 OGSA F2F (19) Dr McGough and Mr Novov attended the OGSA F2F (19). Their main areas of focus were as follows: • OGSA-DMI • OGSA EMS • OGSA Strategy • Info/data modelling architecture • Basic Security Profile Page 126 • Service Level Terms Imperial College, London hosted this OGSAF2F meeting. Imperial were able to provide a venue suitable for the meeting, along with refreshments, thanks to the support from GridNet2. OGSA-DMI The OGSA-DMI session focused on the use of Data EPRs (DEPRs). These would not fit within the current URI elements for source and target within a JSDL document, although because a JSDL source and target has an xsd:any, it could be placed within that. There is also a problem with EPRs, in general, that when they are modified, all the security is lost due to the hashing of the EPR becoming invalid. However, there was space within a DEPR to put security credentials. This was an xsd:any, and although it was possible to just place a security attribute, it was probably better to leave this as such. The consensus was to submit the document for public comment. This work was significant for JSDL as it provided a new and more appropriate way to transfer files and again showed how pervasive JSDL had become in the wider community. This was also significant for GridSAM as it was something that GridSAM was likely to require. OGSA EMS For OGSA EMS scenarios, the question was what to focus on beyond the basic scenarios. Some of the possible contenders for extension are file transfer, advanced filtering, Kerberos security, Info Modelling and accounting. It was proposed to scope just the things that could be done with the HPCB Profile, though some people felt this to be too constraining. There was discussion about file staging and using OGSA-DMI, although there was still concern about DEPRs (see above). Security was also a problem. Because the people running CDDLM were no longer active within OGF and everyone who had evaluated CDDLM felt that it was too complex to implement, it was dropped from the next version of the document. The EMS work was also highly relevant for Dr McGough and Mr Novov’s work on GridSAM and the future development of JSDL. Due to internal reviews held within OGF, there was much refocusing within the OGSA group which led to an OGSA Strategy session. The feeling within the community (from a yet to be published survey for OGF) was that most people in OGF were developers (e-Scientists) and not users, and security was becoming a key problem. At present, no real progress had been made, because over the last four years OGF/OGSA had focused on low level specifications. The group needed to move on from this and move development up through the stack and add in security. There was much discussion about providing higher level APIs that programmers could use (possibly SAGA). There was also discussion of a move towards access layers and service layers and how existing standards can be used together. Further discussion covered the emerging cloud technologies and how these relate to OGF/OGSA. There was a proposal to focus more on the clients – browser (Web 2.0), traditional OS, APIs, APIs for clouds, and mashup technologies. This discussion was highly interesting and something that Dr McGough could contribute towards as his work had moved already towards internet (client) types of technologies. 3 GridNet2 Reports 3.25 Dr Stephen McGough – Imperial The Information/Data Modelling Architecture document is now almost complete. The new section within the document was provided by the JSDL group to define how requirement matching would be provided through future versions of JSDL. This was to be done through the use of XQuery, though it was asked if XPath would be sufficient. This was to be evaluated. The need for optionality, weighting and preference were also discussed and would feed into the next version of JSDL. It was also noted that hierarchies would be needed as, for example, a system responding as a “RedHat” OS should be able to match a job requesting OS of “Unix”. This work was highly relevant for Dr McGough’s work on JSDL as it would feed into the JSDL requirements extension and JSDL 2.0. to a much wider community. As this was a large event, they were able to bring people onto the stand who had seen the HPCP demos on other stands and were keen to know what Dr McGough’s team had achieved. Basic Security Profile Thus, GridSAM would be one of the first full implementations of this OGF-endorsed standard, which should greatly enhance the chances for a wider adoption of GridSAM by the Grid community and easier integration with other Grid-based middleware systems. The Basic Security Profile was now finished and rendered the OGSA Security Profile 1.0 obsolete. There was discussion on who would implement this work. GridSAM was one contender. Service Level Terms For the Service Level Terms work, it was decided that there was not enough consensus to standardise anything as yet and this design team has been left to rest. 3.25.17 SuperComputing ‘06 and OGSA F2F (16) The paramount reason for Dr McGough and Mr Novov to attend this event was to take part in the OGSA HPCP interoperation event at Supercomputing. This event brought the following ten independent implementations of the HPCP profile on JSDL and BES together to demonstrate interoperation in action: • University of Virginia • Microsoft • Platform Computing • Globus • Unicore • Genesis II • OMII-UK/GridSAM • CROWN • gLite WMProxy • gLite CREAM This was a highly significant event for all three groups as it showed not only that the standards had reached a certain level of maturity, but also a significant uptake within the wider community. This was highly relevant work for Dr McGough and Mr Novov, as not only were they providing one of the implementations but it was also the culmination of a significant amount of their effort. They were able to interoperate with all but one of the other projects (which also had significant problems interoperating with others). Dr McGough and Mr Novov were showcasing their software on the UK e-Science stand, which allowed them to present GridSAM, along with the work that they have been doing within the OGF, The success of this interoperability demonstration is likely to influence the future development plans for GridSAM. The required functionality tested at the event comprised only a subset of the full capabilities provided by GridSAM interface. Subsequent development of the HPCP Interface to provide more comprehensive facilities would likely result in morphing of the two and possible future versions of GridSAM offering a full support for the officially standardised HPC Profile. During the OGSA F2F meeting, progress was made with the new approach to resource matching with ClassAds-like documents. Work was made on evaluating XQuery as a language to perform this. Dr McGough and Mr Novov were able to feed into this work through prior experience from GridSAM and their work with ClassAds in the past. There was a discussion about how to extend JSDL towards JSDL 2.0 in the JSDL session. This included a presentation from IBM on how JSDL could be included within a BPEL document to use it as part of workflows. This showed how JSDL was being taken seriously by the wider community. This was also relevant as Dr McGough and Mr Novov had themselves used GridSAM as a service within a BPEL document. 3.25.18 SuperComputing ‘07 Dr McGough and Mr Novov attended SuperComputing 07 (SC07). Their main areas of focus were as follows: • GridSAM • OGSA HPCP interoperability The SC07 convention was again used by the HPCP-WG as a staging ground for a second demonstration of interoperability among a number of prototype implementations of the HPC Basic Profile (HPCBP) v1.0 specification. At the previous year’s event, several groups managed to demonstrate a successful interoperation among different job scheduling platforms through the use of a common interface: the initial draft of the HPCBP. In the year after the event, taking into account the participants' experiences and incorporating viable solutions for the issues encountered, the HPCP-WG refined the Basic Profile and the specification was published as a draft recommendation version 1.0. Relying on the success of the event from the previous year, several engineering teams used the SC07 show floor to set a follow-up demonstration of the improved HPCBP. The second year again brought together a mix of representatives from academia and industry as follows: • Altair Engineering Page 127 3 GridNet2 Reports 3.26 Dr Andrew Martin – Oxford • Microsoft • Platform Computing • NorduGrid • University of Virginia e-Science Group • GridSAM/OMII-UK • EGEE/OMII-Europe. The tested back-end platforms were based on a number of different technologies: • Windows OS • Various version of Linux OS • Java • C++ • C# • Axis • gSOAP • XFire • Windows Compute Cluster • PBS • LSF • SGE • Torque • ARC • CREAM • Globus Despite the heterogeneity, with the use of their respective implementations of the refined HPCBP v1.0, the teams again succeeded in submitting, monitoring, querying and controlling the test suites of computational jobs. What was more, for this second interoperability test, a few of the participants, including GridSAM, University of Virginia, Microsoft and Platform Computing, had enhanced the functionality of their implementations to provide support for two newly proposed extensions of the Basic Profile: • File Staging The File Staging specification, based mainly on the Data Staging element in JSDL v1.0, provided a 'common denominator' protocol for transferring input/output data files as part of a job processing. • Activity Credential The Activity Credential extension provided a standard way of supplying security credentials necessary for a cross-organisational launching of executables and/or staging of files. Out of that smaller group of four, GridSAM was one of the more mature platforms with previous experience with JSDL's Data Staging and security Page 128 credentials processing. That previous experience was invaluable in resolving the various problems occurring during the testing. After the event, each team's experience was documented, along with the workable solutions of the problems encountered. Through the continued work of the HPCP-WG, where GridSAM team had regular participation, the two extension specifications were brought to the public comment phase of the standardisation process. Likewise, the overall success and experience of the SC07 Interop-fest would likely help the HPCBP v1.0 become a full OGF standard. Some time after the event, a couple of the industry participants announced their plans for integration of their HPCBP implementations into their respective upcoming product releases. Knowing for a fact that the GridSAM's HPCBP support had tested successfully with these commercial products, Dr McGough and Mr Novov could now be even more confident about the GridSAM project's potential for its wider acceptance. There had been already concrete plans for the integration of the GridSAM HPCBP-related functionality within the GridBS project. It would be also likely that much of the experience of the GridSAM team gained through involvement with the HPCBP-WG and OGF activities would apply towards the successful completion of the GridBS project. 3.26 Dr Andrew Martin – Oxford Dr Andrew Martin is a lecturer and deputy director of the software engineering programme at the University of Oxford. This report also covers Mr Andrew Cooper’s GridNet2-funded activities. Mr Andrew Cooper is a DPhil student and programme researcher in the software engineering programme at the University of Oxford. For summaries of the GridNet2 activities covered by this award, see the following sections: • Section 2.8, Mr Andrew Cooper – Oxford • Section 2.26, Dr Andrew Martin – Oxford 3.26.1 GGF15 Dr Martin attended GGF15. TC-RG TC-RG began with a BoF at GGF-13, and had been properly inaugurated at GGF-14. This was, then, the first meeting with progress to report. The group was chartered to produce two deliverables, representing the overlapping interests of its two chairs. Martin is the editor of a document recording use cases for trusted computing in Grids contexts, and Wenbo Mao is leading an architecture and implementation project to deliver trusted computing functionality which interoperates, or naturally extends, existing Grid solutions. The discussion at GGF14 had set out some requirements on the use case document; namely, that it should structure the use cases under a small number of headings, relative to the obvious, immediate benefits of the use of TC. Through the group mailing list, other people contributed some of 3 GridNet2 Reports 3.26 Dr Andrew Martin – Oxford these which were already in progress. Since many GGF groups were producing and discussing use cases, their output would also be surveyed for relevant interactions with TC ideas. Use cases would be documented relative to the threat or opportunity they addressed. Little input on the use cases had occurred on the mailing list in the interim, so in preparation for the GGF-15 meeting, Dr Martin had produced a document, based mainly on his own ideas and those of Wenbo Mao, with place-holders for other use cases proposed by interested parties at the original BoF. The document was not ready to be uploaded in advance of GGF-15, so it was described rather than available for advance discussion. The use cases identified were: 1. Server Credential Storage 2. Centralised user credential storage 3. Secure data isolation 4. Process isolation (sandboxing) 5. Trusted information services 6. Distributed firewall; application-aware firewall 7. Secured Audit/Logging Service The discussion indicated that the use cases should represent a spectrum from straightforward easy gains through more innovative examples. This was not because novelty is good per se, but because this technology had the potential to allow big classes of application which had simply been impossible in the past. Genuinely improved functionality (where needed) was worth exploring. resulting document – the first full draft of the use case document – was uploaded to GridForge before Christmas. Dr Martin and Mr Cooper attended GGF-16 to present and discuss this document. As usual, the TC-RG meeting also included a report and subsequent discussion from Wenbo Mao on the progress of his Daonity architecture and implementation. A full architecture draft document was complete in time for GGF-16 and was uploaded to GridForge. The progress of the group was therefore tangible, with two public draft documents. The draft use case document was met with enthusiasm by the relatively small TC-RG meeting. The progress was acknowledged, and the lack of further input regretted. The area directors encouraged the TC-RG to progress the document through the GGF process so that it would perhaps receive further input and comment, and also so that it might reach a wider audience. This was to happen as soon as possible. Ad Hoc BoF on Virtualization It became clear that Trusted Computing alone was too low level a technology to address many application issues by itself. Techniques for system virtualisation (such as Xen, VMWare, VirtualPC, etc.) provided a natural partnering for enhanced trust in applications and middleware infrastructure. The same techniques were also gaining popularity for Grid job management, partitioning of tasks on Campus Grid nodes, and so on. Though it was premature to consider standardisation in this area, it was clearly important that those pursuing different aims through this technology were aware of each other. As a result, Dr Martin organised a BoF on this subject. There was some synergy with the activities of the Firewall Issues Research Group (FI-RG). Leon Gommans had been looking at the Trusted Computing Group's Trusted Network Connect specification and expects to explore this within the FI-RG, since it potentially offered a means for providing ad hoc access through firewalls, as needed for applications like GridFTP. The BoF attracted a good level of attendance, including quite a number of area directors. Speakers included: The meeting also heard a report from Wenbo Mao on progress with the architecture and implementation project. • Andy Cooper, Virtualisation and the Grid • Wenbo Mao, Secure Virtualisation with Trusted Computing • David Wallom, Virtualisation in Campus Grids • Comment by Rhys Francis, who described the Australian Grid virtualised gateway using Xen to run different middleware stacks to support a wide range of users. • Reference was also made to virtualisation extensions to Globus Toolkit 4, which Ian Foster had described in his keynote address. • Comment by Matt Viljoen, who described a Rutherford Lab process for distributing client Grid tools using a VM player. Other Sessions During GGF15, Dr Martin also attended and contributed to meetings of the FI-RG, the OGSA-D Security Discussion (wherein several use cases arose which would be directly addressed by TC technologies), the OGSA-Authz-WG, and a community session on Leveraging Site Infrastructure for Multi-Site Grids. 3.26.2 GGF16 Dr Martin and Mr Cooper attended GGF16. TC-RG At GFF-15, an outline for a use case document was presented and discussed. A little further discussion had occurred on the mailing list since then, and Mr Cooper, who had been named on the original GridNet2 funding proposal, contributed substantially to the document in the last months of 2005. The There was a good level of discussion, concluding that two threads of activity should be taken forward: • Consideration (eventually towards standardisation) of how virtualisation technologies can be integrated with Grid middleware to achieve the desirable outcomes • Community means of keeping in contact those using these technologies in production Grids. Page 129 3 GridNet2 Reports 3.27 Dr Steven Newhouse – OMII-UK These will lead to further BoFs at future GGFs, in the first instance. beginning of an ongoing low-level engagement with these activities. Shibboleth and Grid Note An extended community activity reviewed existing and possible projects integrating Shibboleth and Grid technologies. Dr Martin attended and gave a presentation on the ShibGrid project. In parallel with Dr Martin’s booking and attending this TCG meeting, the OGF area directors decided that the TC-RG should be wound up, due to lack of broad participation. Dr Martin expects the trusted computing agenda to be pursued through the newlyformed research group on virtualisation (GRIDVIRTRG), which he helped to promote by running a BoF at GGF16. Other Sessions Dr Martin and Mr Cooper attended other relevant sessions at the GGF-15 meeting, including acting as note-taker at the Authz interoperability meeting. 3.26.3 Trusted Computing Group Meeting Dr Martin attended the members meeting for the Trusted Computing Group held on 14-16 November 2006 in San Antonio, Texas. Preamble Dr Martin’s GridNet2 proposal mainly covered attendance at GGF meetings in order to advance the work of the TC-RG of which he is co-chair. The Trusted Computing Group (TCG) is the standards body for Trusted Computing technologies, and runs along similar lines to OGF, albeit with a closed membership. Having obtained membership, Dr Martin proposed that it would be helpful to attend a TCG meeting in order to advance the goals of TC-RG, in particular to share ideas and solicit contributions. On this occasion, a time-slot became available which allowed him to attend. Sessions Dr Martin attended the three-day meeting in San Antonio. In particular, he attended a plenary tutorial session, the opening plenary, with news of a TCG collaboration with DMTF (which also collaborates with OGF), and plenaries on Trusted Computing Authentication technologies and protocols. He also attended working group meetings on virtualisation, authentication, server issues, and a joint working group on enterprise storage and key management. Each of those work group areas had a small intersection with the work of TC-RG. Most of their focus is on detailed low-level work which is outside the scope of the Grid middleware he had been considering. As such, he did not expect to be able to contribute very much to their detailed work. It became clear, however, that the Grid requirements constitute a significantly different angle from most of those already being considered. As such, he expects to be able to contribute TC-RG Grid use cases to most of those work groups, and this could be quite a constructive way to encourage future releases of technology to support Grid scenarios. Dr Martin also held one-on-one chats with representatives of Hewlett-Packard, Microsoft, CESG, The German Federal Information Ministry, Seagate, Dartmouth College, and others, explaining his interest in trusted Grid technologies. There was much warm support – and surprise that TC-RG has not attracted more active participants – but no direct likelihood of additional input through OGF. In summary, he judged his attendance to have been worthwhile for the goals of GridNet2, and the Page 130 3.27 Dr Steven Newhouse – OMIIUK Dr Steven Newhouse was the director of OMII-UK, based at the University of Southampton. Dr Newhouse did not use his entire GridNet2 award before leaving OMII-UK, so his replacement, Mr Neil Chue Hong, used the remainder in accordance with the terms of Dr Newhouse’s original GridNet2 application. For a summary of Dr Newhouse’s GridNet2 activities, see section 2.27, Dr Steven Newhouse – OMII-UK. 3.27.1 GFSG F2F – Jan 06 The GFSG F2F meeting took place away from a GGF meeting to allow more time for discussion and planning as to the future of GGF. In addition to the regular business of planning future events and directions, the major upcoming issue was the proposed merger of GGF and the EGA. Dr Newhouse’s primary role in these meetings was to represent the groups within the Applications Standards area of GGF where he was one of the two Area Directors. His primary concern at this meeting was to ensure that the new organisation would remain focused on the standards activity and that the merger would have minimal disruption to this work. 3.27.2 OGSA F2F – Jan 06 To make progress on the standards work, it was necessary to meet between the main GGF meetings. The OGSA activity, which comprises several working groups, met frequently between GGFs to discuss technical issues in depth and to make further progress on standards. Dr Newhouse was actively engaged in the following segments of the meeting: • Basic Execution Service There was a review and discussion of the current draft specification. • Execution Management Services There was continued brain storming and definition of the incremental scenarios that would be used to drive the definition of service interfaces and their interactions in this part of the OGSA model. • Information Model There was a critical discussion, from the perspective of job submission aspects, about building a Grid infrastructure. The group further refined the proposed CIM-based model so that 3 GridNet2 Reports 3.27 Dr Steven Newhouse – OMII-UK the attributes focused on those relevant to the e-Science community as opposed to device manufacturers. 3.27.3 GGF16 within OMII-UK and the managed programme. There were also several closed meetings with the OMII-UK management team. 3.27.4 GGF17 Grid Forum Steering Group Meeting Grid Forum Steering Group Meeting There was an all-day meeting of the GGF Area Directors and the GGF leadership to review and discuss internal activity. This meeting was primarily concerned with the progress of the merger with the Enterprise Grid Alliance. There was an all day meeting of the GGF Area Directors and the GGF leadership to review and discuss internal activity. This meeting was primarily concerned with the continued progress of the merger with the Enterprise Grid Alliance. Grid Interoperability Now High Performance Computing Profile BoF Following on from its initial meeting at Supercomputing 2006 in Seattle, the GIN-CG held a day-long workshop at GGF16. This meeting reviewed the areas that had been set up following the initial meeting: Jobs, Authorisation, and Data and Information Services. This was the first opportunity to have a F2F meeting to review progress. As the coordinator of activity in the Jobs area, Dr Newhouse provided an update of progress to date and future plans to build islands of interoperation around pre-WS-GRAM, WS-GRAM, and BES & JSDL. OGSA Round Table The OGSA-WG was planning its future strategy, looking at which specifications to use to build the OGSA architecture. During a round table session, Dr Newhouse presented a strategy based on the pragmatic use of web service specifications that have been widely adopted. This was based upon the needs of both OMII-UK and the UK e-Science programme. Simple API for Grid Applications Dr Newhouse reviewed the document and contributed to the discussions relating to the document that took place over several sessions. Interoperability Fests BoF One of the issues that was starting to face GGF was how to verify implementations relating to the specifications being developed by the working groups. Various options were discussed but no concrete actions came out of the meeting as no-one present was willing to lead the activity. Basic Execution Service Working Group This proposed activity built on two established activities within the GGF – the Job Submission Description Language (JSDL) and the Basic Execution Service (BES). As groups started to examine how to build on top of these specifications, it was observed that additional constraints needed to be defined, for example, security. A profile on these specifications, that is, the set of additional constraints needed for these specifications to be implemented, had to be defined and this proposed working group would undertake the work. Astro-RG Workshop This workshop aimed to build links between GGF Standards groups and the domain specific specifications being set in the Astronomy domain through the International Virtual Observatory Alliance. Dr Newhouse presented ongoing work in the Applications Standards area (primarily the SAGA activity) and how that was driving the development of APIs that could protect application scientists from changing services and interfaces when adopting Grid infrastructures. OGSA Execution Management Services An activity that Dr Newhouse had initiated at the GGF15 meeting was to develop concrete use cases to drive the specification of services in the EMS space. This involved a review of the current document and use cases. Feedback from the session identified where use cases needed to be clarified and where new use cases where needed. The document would continue to be developed between GGF meetings. Simple API for Grid Applications As co-chair of the working group, Dr Newhouse led a review of the latest draft of the specification and the discussions that followed. Dr Newhouse reviewed the document and contributed to the discussions relating to the document that took place over several sessions. Applications Area Meeting Bridging the Divide: Community Application Requirements Driving Standards Development As one of the Area Directors for Applications Standards, Dr Newhouse led this meeting that reviewed activity taking place in the working and research groups active in the area. The group decided to plan a workshop for the next meeting that would strengthen the links between activity in the Applications area and other standards activity. OMII-UK Dr Newhouse ran a workshop describing the software components being produced by the OMII-UK consortium. Presentations were provided by partners Dr Newhouse co-organised this workshop which was designed to show case activities within the Application Standards area to other working groups in the standards functions and the broader GGF community. The reason for this was to show that some of the OGF API specifications were well on their way to becoming standards and groups should consider adopting them and contributing to the activity by providing feedback. Page 131 3 GridNet2 Reports 3.28 Mr Vesselin Novov – Imperial Web Services Standards Convergence BoF The partitioning of the mainstream commercial web services standards community from the Grid community through the OGSI and WSRF family of specifications had produced considerable polarisation within the community. Work between the major commercial web services organisation had produced a new family of specifications, based around WS-Management, which would unify these two communities. Collectively they are called WS Resource Transfer. This meeting provided a limited update from the people involved in the detailed negotiations which where currently confidential. In addition to his active involvement and contribution to the above working groups, he also had the opportunity to establish and maintain relationships with other people in the Grid community, including Chris Smith (Platform Computing), Jon MacLaren (Louisiana State University), Tony Hey (Microsoft), Marvin Theimer (Microsoft), Geoffrey Fox (Indiana University), Dennis Gannon (Indiana University), Marty Humphrey (University of Virginia), Steven Newhouse (OMII), and many others. 3.30 Dr Colin Perkins – Glasgow Grid Interoperability Now Dr Colin Perkins is a lecturer in the Department of Computing Science in the University of Glasgow. The GIN-CG held a set of public and private planning meetings dealing with building interoperability between different Grid infrastructures. For a summary of Dr Perkins GridNet2 activities, see section 2.30, Dr Colin Perkins – Glasgow. Applications Area Meeting As one of the Area Directors for Applications Standards, Dr Newhouse led this meeting that reviewed activity taking place in the working and research groups active in the area. The group reviewed the success of the workshop and discussed future activities in the area. 3.28 Mr Vesselin Novov – Imperial The reports for Mr Vesselin Novov are consolidated in Dr Stephen McGough’s reports. For more information, see section 3.25, Dr Stephen McGough – Imperial. For a summary of Mr Novov’s GridNet2 activities, see section 2.28, Mr Vesselin Novov – Imperial. 3.29 Dr Savas Parastatidis – Newcastle Dr Savas Parastatidis was the Chief Software Architect at the NEReSC, School of Computing Science, University of Newcastle. Dr Perkins stated that the GridNet2 funding was essential for his work, and believes the outcomes of this work would be of interest to the e-Science community, and would underpin future developments in Grid toolkits and collaborative work environments. 3.30.1 IETF64 Dr Perkins attended the 64th IETF meeting, held in Vancouver, 6-11 November 2005. GridNet2 funded Dr Perkins’ continued activities as co-chair of the Audio/Video Transport (AVT) and the Multiparty Multimedia Session Control (MMUSIC) working groups in the IETF, and his work to improve the quality and flexibility of protocols for collaborative work and other real-time applications. Opinions expressed herein are those of Dr Perkins, and do not necessarily represent those of other members of the IETF or its working groups. The AVT working group develops standards for realtime audio/visual data transfer over IP networks. Those standards underpin modern voice over IP (VoIP) and video conferencing systems, including the AccessGrid. Work in AVT was focused on: • Management tools for large scale deployment of networked collaborative environments For a summary of Dr Parastatidis’ GridNet2 activities, see section 2.29, Dr Savas Parastatidis - Newcastl. • Multiple description coding for scalability and loss tolerance 3.29.1 • Codec control for loss tolerance • Support for new audio/visual coding standards GGF14 Dr Parastatidis attended GGF14 as a representative of the North-East Regional e- Science Centre (NEReSC). GGF meetings bring together those interested in defining the next generation of standards for building Grid applications. As a result, participation in working group meetings involves technical discussion and exchange of ideas in the areas of distributed computing. Dr Parastatidis was able to contribute his ideas in the areas of service-oriented, distributed computing and in the technology area of Web Services. Dr Parastatidis participated in the working group meetings of the OGSA, the BES, and the DAIS working groups. He is one of the contributors to the DAIS specification. At the meeting, he contributed ideas towards the design of the DAIS protocols. Page 132 During the trip, Dr Perkins spent considerable time working with the open source community to develop RTP payload formats for Speex, Vorbis, and Theora codecs (which were reasonably high quality and were claimed IPR-free, hence being good candidates for inclusion in future AccessGrid versions). Dr Perkins also spent some time working with the authors of the new scalable video coding proposal (an extension to H.264) to ensure it fitted within the RTP framework; this could enable future rate-adaptive conferencing, extending the reach of networked video conferencing to more heterogeneous environments. The RTP Payload Format for uncompressed video, of which Dr Perkins was co-author, was published shortly before this meeting, and an extension to that format was approved at the meeting; these would likely prove 3 GridNet2 Reports 3.30 Dr Colin Perkins – Glasgow useful for high quality interactive networked visualisation applications. • The MMUSIC working group develops protocols for initiation and control of real-time and audio/visual sessions. The following were the primary areas of work in MMUSIC: The work in MMUSIC was concentrated on the following: • • New approaches to negotiating secure media streams • Interactive Connectivity Establishment (ICE) methodology for NAT traversal Interactive Connectivity Establishment (ICE) methodology for NAT traversal, • Development of extensions to the Session Description Protocol (SDP) Development of extensions to the Session Description Protocol (SDP) • Signalling aspects of the negotiation of secure media streams. Both ICE and SDP were essential to future networked virtual environments, forming the infrastructure on which one builds other session initial protocols. Most activity in the MMUSIC sessions Dr Perkins chaired at the 64th IETF meeting related to ICE and its interactions with SIP PRACK and UPDATE methods. Dr Perkins also worked with the Area Directors to speed approval of various SDP extensions, and the newly revised SDP specification of which Dr Perkins is the document editor. More details of the work done in the AVT and MMUSIC working groups can be found in the official proceedings of the 64th IETF meeting. Dr Perkins prepared the minutes for these working groups. For more details, see the following: • http://www.ietf.org/proceedings/05nov/avt.htm l • http://www.ietf.org/proceedings/05nov/mmusic .html In addition to his work chairing the AVT and MMUSIC working groups, Dr Perkins presented a new proposal for transport of real-time audio/visual data (carried in RTP) over the newly standardised Datagram Congestion Control Protocol (DCCP). DCCP was an interesting new network transport protocol, well suited for traffic that prefers timely delivery to complete reliability, yet elastic enough to accept congestion control. His work in this area was initially focused on streaming audio and video over DCCP, and should eventually lead to improved quality and a more flexible service compared to current “HTTP-streaming” solutions. In the future, DCCP could prove an appropriate transport for all kinds of real-time data, for example feeds from sensor networks, and RTP over DCCP was an appropriate protocol base to host such applications. 3.30.2 IETF65 Dr Perkins attended the 65th IETF meeting, held in Dallas on 19-24 March 2006, in his continuing roles and activities as co-chair of the Audio/Video Transport (AVT) and the Multiparty Multimedia Session Control (MMUSIC) working groups in the IETF, as well as his work to improve the quality and flexibility of protocols for collaborative work and other real-time applications. The work in the AVT group was focused on the following: • Management of large scale VoIP deployments through updates to the RTP management information base (MIB)and the RTP Control Protocol Extended Reports • Development of RTP payload formats to support new media codecs During the trip, Dr Perkins co-chaired sessions of both the AVT and MMUSIC working groups, and conducted numerous ad-hoc meetings with document authors to discuss their drafts and resolve open issues. Dr Perkins contributed to the following key areas: • Definition of signalling protocols for layered media coding to enable adaptive large group conferences with the new H.264 scalable video coding (SVC) extensions – potentially directly applicable to the AccessGrid • Discussion of the potential harmonisation of the Session Initiation Protocol (SIP) and the RealTime Streaming Protocol (RTSP) to enable converged services • Discussion on the formation of a new working group to study forward error correction codes for streaming media and other real-time sessions (potentially useful to data streaming e-science applications such as eVLBI) • Discussion of the RTP management framework (again, useful to better diagnose problems in AccessGrid sessions). There was also considerable ongoing work to improve the signalling for secure RTP sessions, of which the ZRTP proposal was the best known of several alternatives. This was of huge importance to the emerging VoIP market, but also affected other networked collaborative environments. For more details of this and the other work done in the AVT and MMUSIC working groups, see the official minutes of the 65th IETF meeting, which Dr Perkins prepared in part: • http://www.ietf.org/proceedings/06mar/minute s/avt.txt • http://www.ietf.org/proceedings/06mar/minute s/mmusic.txt In addition to chairing the AVT and MMUSIC working groups, Dr Perkins presented an update to his proposal for transport of RTP data over the Datagram Congestion Control Protocol (DCCP). DCCP is a protocol that is gathering interest in the e-Science community since it allows for a range of congestion control responses, making it better suited to the needs of large scale streaming data applications than TCP or UDP. Dr Perkins’ work brings RTP framing to DCCP, making it ideally suited to both video streaming and visualisation, and to bulk real-time data streaming (for example, sensor data or eVLBI). Page 133 3 GridNet2 Reports 3.30 Dr Colin Perkins – Glasgow 3.30.3 IETF66 3.30.4 IETF67 Dr Perkins attended the 66th IETF meeting in held in Montréal on 10-14 July 2006 in his continuing roles and activities as co-chair of the Audio/Video Transport (AVT) and the Multiparty Multimedia Session Control (MMUSIC) working groups in the IETF, and his work to improve the quality and flexibility of protocols for collaborative work and other real-time applications. Dr Perkins attended the 67th IETF meeting, held in San Diego from 5–10 November 2006, in his continuing roles and activities as co-chair of the Audio/Video Transport (AVT) and Multiparty Multimedia Session Control (MMUSIC) working groups in the IETF, and his work to improve the quality and flexibility of protocols for collaborative work and other real-time applications. The AVT working group develops standards that underpin modern voice over IP, 3G mobile telephony, and video conferencing systems, including the AccessGrid, VRVS, and similar tools used by the escience community. Dr Perkins co-chaired the AVT working group session at the 66th IETF meeting, and conducted numerous ad-hoc meetings with document authors to discuss their work and to help resolve open issues in a timely manner. The main areas of discussion in the working group are as follows: Dr Perkins co-chaired the AVT working group session at the 67th IETF meeting. The main area of discussion in the working group session and in several ad-hoc meetings focused on the following: • Video codec control messages, to enhance support for centralised multiparty conferences, as a robust alternative to the multicast architecture used by the AccessGrid The group came to general agreement that the current DTLS-based proposal was a better fit to the RTP architecture than the ZRTP proposal, and suggested several approaches by which ZRTP could be adapted. • Definition of the RTP Payload Format and signalling for the H.264 scalable video coding (SVC) extensions to support extremely large scale heterogeneous conferencing sessions • Improved diagnostic and management tools The major current work item in MMUSIC is the Interactive Connectivity Establishment (ICE) methodology for NAT traversal, which is needed to allow effective use of signalled real-time media in the presence of NAT devices. This is directly applicable to collaborative work tools, such as AccessGrid, but potentially has much wider use, since the mechanism is general enough to be used for NAT traversal for file transfer, RPC, and other signalled sessions. As such, the underlying technology can be used by Grid Computing and Peer-to-peer applications, as well as by voice-over-IP and video conferencing systems. Dr Perkins chaired the MMUSIC session at the 66th IETF meeting, managing discussion of these issues. There is also considerable ongoing work in the IETF to enhance the security of voice-over-IP and other video conferencing systems. Dr Perkins is closely engaged with this work to ensure the resulting standards sit well with the RTP protocol architecture, and do not ignore the needs of the large scale conferencing and e-Science communities (both of which are sadly under-represented in the IETF). This work took place in the RTPSEC BoF and mailing list. In addition to his work chairing these working groups, Dr Perkins contributed to the Datagram Congestion Control Protocol (DCCP) working group. Dr Perkins presented an update to the specification for transport of RTP-over-DCCP, which was accepted as a work item of the DCCP group at this past meeting, and will form the basis for future video streaming and visualisation systems, and also bulk real-time data (for example, sensor or eVLBI) streaming. Dr Perkins also presented preliminary results to show performance of applications using TFRC Friendly Rate Control (as used in DCCP CCID 4) when running over real-world networks. This latter presentation was well received by the community of DCCP implementers. Page 134 • Secure negotiation of encryption and authentication keys for multimedia sessions • Consideration of how the ZRTP and DTLS-based keying proposals can be adapted to better match the RTP architectural model In addition to chairing the session, Dr Perkins presented two pieces of work to the AVT working group. The first of these was new work on the mechanisms and signalling needed to multiplex RTP and RTCP on a single UDP port to ease NAT traversal. This was very well received, with widespread interest from the working group, and would be developed over the coming months. The second piece of work was on use of RTP with the Datagram Congestion Control Protocol (DCCP) to help make multimedia applications adaptive to network congestion. This was presented in both AVT and the DCCP working groups to get wide review as it neared completion. It was expected that this document would go to working group last call before the end of the year, and be completed early in 2007. Dr Perkins co-chaired the MMUSIC session at the 67th IETF meeting, managing the discussion office methodology for NAT traversal, and was involved in several related discussions. ICE was expected to be completed in the first half of 2007, and Dr Perkins continued to be closely involved in managing the publication of this standard. In addition to his work chairing and contributing to these working groups, Dr Perkins participated in a meeting with Marshall Eubanks (representing part of the eVLBI community) on how to develop effective network transport protocols for eVLBI data. This followed on from a presentation Marshall gave to the IETF Transport Area, and made use of Dr Perkins’ past experience with transport of uncompressed high definition video, to discuss how to use RTP as a transport protocol for eVLBI streams. In the course of this meeting, Dr Perkins and Eubank identified several problems with the current proposal for eVLBI streaming which could lead to loss of time synchronisation, and began initial discussion of alternative proposals. 3.30.5 IETF68 Dr Perkins attended the 68th IETF meeting in held in Prague from 18–23 March 2007 in his continuing role and activities as co-chair of the Audio/Video Transport (AVT) and Multiparty Multimedia Session 3 GridNet2 Reports 3.30 Dr Colin Perkins – Glasgow Control (MMUSIC) working groups in the IETF, and his work to develop protocols for collaborative work and other real-time applications. Dr Perkins co-chaired the AVT working group session at the 68th IETF meeting. The main areas of discussion were as follows: • • • Generalisation of the RTCP framework to allow non-compound packets and to report more detailed performance metrics for IPTV systems Extensions to RTP to support and signal layered codecs, in particular the SVC extension to H.264; Codec control messages, to better support centralised small group video conferencing systems Congestion control for real-time streaming sessions. This work was primarily aimed at improving the quality of small group video conferencing systems, although the latter item was also applicable to data streaming (for example, the eVLBI work discussed later). The work that Dr Perkins presented at the previous IETF meeting on multiplexing RTP and RTCP on a single port to ease NAT traversal went to working group last call at the meeting, and was expected to be published as a Proposed Standard in the coming months. His work on RTP and the Datagram Congestion Control Protocol (complementing the RTP congestion control work) was discussed in the DCCP working group. Issues with keep-alives, to ensure NAT pinholes remain open, were discussed. The draft was expected to go to working group last call shortly, once the RTP and RTCP multiplexing work, on which it depends, was completed. Dr Perkins co-chaired the MMUSIC session at the 68th IETF meeting, managing the discussions of ICE methodology and the SDP media capability negotiation framework, and related issues. Dr Perkins also participated in the design team developing the media capability negotiation framework, with a particular emphasis on deployability and backwards compatibility of the resulting protocols, ensuring they fitted with the SDP architecture. (Dr Perkins was one of the authors of the SDP specification.) In addition to his work chairing and contributing to these working groups, Dr Perkins continued discussions with Marshall Eubanks (representing part of the eVLBI community) on developing effective network transport protocols for eVLBI data, utilizing work done in the Audio/Video Transport working group as a basis. This followed on from previous discussions at the 67th IETF meeting. 3.30.6 IETF69 Dr Perkins attended the 69th IETF meeting in held in Chicago from 21–28 July 2007 in his continuing roles and activities as co-chair of the Audio/Video Transport (AVT) and Multiparty Multimedia Session Control (MMUSIC) working groups in the IETF, and his work developing standards for real-time collaborative work applications. Dr Perkins co-chaired both AVT working group sessions at the 69thIETF meeting. Discussion focused on the following issues: • Securing RTP sessions • Management of RTP session, including extended reception quality feedback • Use of non-compound RTCP to reduce the overhead of congestion control and reception quality feedback, particularly on wireless links • Keep-alive mechanisms for NAT traversal • Various new RTP payload formats, including scalable video coding The main aims of this work are to improve security and quality of multimedia communication. Dr Perkins co-chaired both MMUSIC sessions at the 69th IETF meeting. Both the ICE Methodology for NAT traversal and SDP media capability negotiation frameworks were essentially complete. ICE depended on Perkins’ work multiplexing RTP and RTCP on a single port to reduce connection setup times. Dr Perkins’ work on transporting RTP sessions over the Datagram Congestion Control Protocol (DCCP) had been approved for publication as a Proposed Standard RFC, subject to completion of on draft on multiplexing RTP and RTCP on a single port. That draft was in IETF-wide “last call” for comments, and was expected to be published as a proposed standard RFC in the coming months. There had been some discussion relating to IANA registration of new RTCP packet types, inspired by this draft, which were clarified with IANA. Both drafts were discussed at the meeting, both in their own right, and in the context of ICE, and reducing the overhead of DTLS security negotiation for RTP sessions. There was now considerable interest in this work, especially the multiplexing draft, to reduce RTP connection overheads. 3.30.7 IETF70 Dr Perkins attended the 70th Internet Engineering Task Force (IETF) meeting held in Vancouver from 27 December 2007, in his role as co-chair of the Audio/Video Transport (AVT) working group, and his work developing standards for real-time collaborative work applications and Network Address Translator (NAT) traversal. Dr Perkins co-chaired both AVT working group sessions at the 70th IETF meeting. The discussion in the first session focused on the following: • RTP payload formats for layered audio and for scalable, multi-view and 3-dimensional video • Reporting multimedia quality metrics for performance monitoring and systems management Dr Perkins was actively involved in these discussions, and in several side meetings in these areas, working to ensure industry proposals are coherent and aligned with the RTP architectural framework. The second AVT session focused on the following: • Use of non-compound RTCP (a draft from Ericsson building on Perkins’ previous work on multiplexed RTCP) • Security of RTP sessions Page 135 3 GridNet2 Reports 3.31 Dr Stephen Pickles – Manchester The security work was gaining large interest, and should provide an infrastructureless mechanism for keying secure RTP sessions. Dr Perkins was again strongly involved, helping to integrate the security mechanisms (Diffie-Helman key exchange) into the RTP framework. Dr Perkins also chaired a BoF session on Self Address Fixing Evolution. One of the key problems in deploying peer-to-peer systems in the current Internet was NAT traversal. This affects multimedia conferencing, telepresence, and VoIP systems – which was his main interest in this work – but also directly affected Grid computing systems which must transfer files and data or execute jobs across organisational boundaries. The purpose of this BoF was to explore the limitations of existing protocols in this area, and to gauge interest in possibly chartering future work in the IETF to resolve some of the issues raised. Discussion was lively, and provided a great deal of input to the area directors. Finally, Dr Perkins continued to participate in the Transport Area Directorate of the IETF, advising the area directors on transport aspects of real-time media protocols. 3.30.8 SUMOVER – Nov 05 Dr Perkins attended the SUMOVER workshop in London, 28-30 November 2005. The SUMOVER project was a JISC-funded development activity at University College London (UCL) that aims to provide for support and development of the Mbone Videoconferencing tools for the research community. It was a follow-on to the early development work at UCL that produced the Robust Audio Tool (rat) and continued development of the vic video conferencing system, the tools which underpin the AccessGrid virtual collaborative environment. Dr Perkins was invited to attend the SUMOVER workshop for several reasons: 1. Several years ago, Dr Perkins was one of the original developers of rat and has extensive experience with that system. 2. Dr Perkins’ research into HDTV over IP and congestion controlled and peer-to-peer algorithms for video conferencing was directly relevant to SUMOVER. 3. Dr Perkins co-chaired the IETF Audio/Video Transport (AVT) and Multiparty Multimedia Session Control (MMUSIC) working groups which develop related standards. Dr Perkins’ participation in the SUMOVER workshop was two-fold: • He discussed his research results where relevant to the aims of SUMOVER. • He discussed how new developments in the IETF community might affect the development of the AccessGrid and hence the tools maintained by the SUMOVER project. • A heavy push for NAT traversal solutions, such as the ICE methodology being developed in the IETF MMUSIC working group. • Development of new RTP profiles for security and to provide improved feedback for reception quality monitoring and codec control. • Development of new RTP payload formats for iLBC, AMR, H.264 and their variants. • Development of new congestion controlled transport protocols for audio/visual media (for example, the RTP profile for TFRC, and DCCP). • Continued push to deprecate traditional IP multicast and replace it with a Source Specific Multicast system; work on QoS; peer-to-peer algorithms; E911. Dr Perkins was the author of the updated SDP specification and the new specification for RTP over DCCP, and the SUMOVER workshop offered an invaluable opportunity to socialise this specification prior to its adoption by the IETF. It also provided a valuable opportunity for him to seek input from the open source and academic/e-Science communities on their priorities, which he was able to take back to the IETF through his position as working group chair. 3.30.9 SUMOVER – Apr 07 Dr Perkins attended the SUMOVER workshop held at UCL on 26 April 2007. Dr Perkins attended the SUMOVER workshop to provide advice on the use of IETF standards within the SUMOVER project. As such, it directly followed from his attendance at IETF meetings, which form the major part of his GridNet2 activities. In particular, the SUMOVER project was integrating several new video codecs into the AccessGrid system. These included the H.264 codec, which would give significant improvements to the quality and robustness of AccessGrid video in future releases of the toolkit. The codec was integrated using the RTP payload format developed in the IETF audio/video transport working group, of which Dr Perkins was cochair, and he had provided extensive input to the design of the format. He gave input to the project on the best way to use this standard, on the intellectual property issues relating to it, and on the extensions to the standard that were under development in the IETF, which would affect SUMOVER. Dr Perkins also provided the project with an overview of other developments in the IETF standards community, which might affect their development work. These included the RTPSEC security work (the competing ZRTP and DTLS pre-drafts), NAT traversal, and congestion control. Unlike the H.264work, which was currently being integrated into the AccessGrid toolkit, these subjects were of primarily longer term interest, for future releases. 3.31 Dr Stephen Pickles – Manchester The developments in the IETF that may be relevant include: Dr Stephen Pickles was Team Leader for Grid Developments at the University of Manchester. • For a summary of Dr Pickles’ GridNet2 activities, see section 2.31, Dr Stephen Pickles – Manchester. Continued development of SIP-based solutions for session initiation and control, using the SDP offer/answer model for media negotiation. Page 136 3 GridNet2 Reports 3.31 Dr Stephen Pickles – Manchester This report covers Dr Pickles’ activities in SDOs that have been facilitated by GridNet2 funding. These were mostly within the OGF (formerly the GGF), but also OASIS and ETSI. 3.31.5 3.31.1 The Multi-Grid Interoperation initiative started by Charlie Catlett, Satoshi Matsuoka Satoshi, Ray Bair, Dane Skow and Dr Pickles in September 2005 became the GIN-CG. GIN held a well-attended workshop, composed of four sessions, and the major production Grids signed up. OGF Dr Pickles was elected to the GFSG in September 2004, as Area Director for what was then the Scheduling and Resource Management (SRM) Area. Dr Pickles stood down as co-chair of the SAGA Research Group (which he helped to found) shortly after. During the period of this report, the OGF Areas were re-organised, and the SRM Area was effectively split into two areas, that is, the Compute Area and the Management Area. Dr Pickles continued as AD of the Compute Area, but also remained active in the accounting working groups (UR-WG and RUS-WG) which moved to the Management Area. This re-organisation was overshadowed by the merger of GGF and the Enterprise Grid Alliance to become the OGF, and this topic dominated the discussions in GFSG. Nonetheless, the OGF managed to keep the specifications flowing; indeed the pace seemed to pick up a little. During this period, Dr Pickles was the author of three published OGF documents, an acknowledged contributor on four, and influenced or shepherded numerous others. The Compute Area made great progress with JSDL, OGSA-BES, and OGSA-HPC-Profile, and WSAgreement (GRAAP-WG) finally became a proposed recommendation. Dr Pickles also helped to start the GIN activity. 3.31.2 OASIS The WSRF specifications were ratified as OASIS standards on 24 April 2006. Dr Pickles co-authored an HPDC paper (Humphrey, Foster et al.) on WSRF interoperability, which helped to restore community confidence in WSRF. Dr Pickles’ RA, Mark McKeown, contributed to the development of these standards and provided an implementation of them in Perl. Dr Pickles also participated in some calls of the OASIS WSRF Technical Committee, including the last. 3.31.3 ETSI ETSI (http://www.etsi.org/) is a long-established European SDO with roots in telecommunications and interoperability testing, and has won European funding to address topics in the Grid space. Dr Pickles took on a temporary liaison function between OGF and ETSI, and represented OGF at the third meeting of the ETSI Technical Committee on GRID (TC-GRID). 3.31.4 GGF15 The SRM Area was split into two areas, Compute and Management. Dr Pickles’ co-AD for SRM, Bill Nitzberg, did not stand for another term and Ramin Yahyapour joined him as co-AD for the Compute Area. Hiro Kishimoto and Ellen Stokes took over the Management Area. Dr Pickles continued to take a special interest in the accounting groups UR-WG and RUS-WG, which have been moved to the Management Area. GGF16 The GFSG worked on the details of new organisation arising from the merger of EGA and GGF. Dr Pickles also organised a BoF on interoperability, in the context of bringing GGF specifications to full recommendation status. This meeting, and followups, led to the establishment of the good practice of writing an informational document on plans for interoperability testing prior to actually conducting the tests. The DAIS-WG was the first working group to do it this way, on Dr Pickles’ advice. 3.31.6 GGF17 The GridWorld Japan attracted a tremendous 3000 participants. Dr Pickles was busy with working groups. 3.31.7 GGF18 GGF18 was held in conjunction with GridWorld; this was not very successful, and IDG and OGF decide not to continue their co-location agreement. The GFSG was strengthened by new blood from EGA. Manchester was announced as host city for OGF20. 3.31.8 OGF19 This was a five-day meeting with a synchronisation day mid-week (following the W3C model) and was pronounced successful by all. The numbers were down, but much good business was accomplished. Security moved up the priority list of many OGF working groups. A BoF on Levels of Assurance attracted a remarkable level of interest; in the months following it became a new research group. 3.31.9 OGF20 Dr Pickles acted as chair of the Local Organising Committee (LOC) for OGF20 and the 2nd EGEE User Forum. This was complicated by the number of organisations involved: OGF (the customer), EGEE (another customer), UK e-Science (the formal host), and the University of Manchester (LOC). Manchester Central, the venue, underwent re-branding and staff turnover, and numerous other committees and colocated events that were hosted by the University were also involved. In the end, the events were successful, thanks to sustained efforts by many. Dr Pickles also shared the organisational responsibilities of a BoF (Workshop on Heuristics for Implementing Semantic Knowledge Yardsticks), and a two-session workshop on Computational Steering. Dr Pickles was not able to attend all of OGF20, due to illness, but was able to attend the GFSG steering meeting. 3.31.10 GFSG F2F – Jan 06 Dr Pickles attended the GFSG F2F meeting held in San Francisco, on 11-13 January, 2006. The discussion at this meeting focused on the forthcoming merger between GGF and the Enterprise Grid Alliance. Page 137 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff 3.31.11 GFSG F2F – Jul 06 1. Frameworks and Approaches Dr Pickles attended the GFSG Standards Council F2F meeting, held in London on 26-27 July 2006. 2. Use Cases/Applications and Combining Technologies Fujitsu Laboratories Europe hosted this meeting, attended by GFSG members in the Standards area. During this meeting, the GFSG members set the standard priorities for the new merged organisation, refined the standards roadmap, and formulated strategies to improve the document process. 3. Technologies and Standards 3.31.12 GFSG F2F – Dec 06 Dr Pickles attended the GFSG F2F meeting at held e-Science 2006, Amsterdam, on 6-8 December 2006. This two day F2F meeting followed the OGF session held at e-Science 2006. 3.31.13 OGF GIN F2F – Mar 07 Dr Pickles attended the OGF GIN F2F Meeting, held at CERN on 14 March, 2007. Semantic Grids: Frameworks and Approaches A key presentation in the first session focused on the developed of “S-OGSA” – essentially an approach to providing semantic description of services within OGSA. Such an approach was intended to provide the ability to add semantic content to each service that was made available over existing Grid systems (such as Globus), whilst still maintaining compatibility with OGSA. Using this approach, metadata stores and ontology engines were essentially modelled as data services in OGSA, thereby enabling the use of OGSADAI and related data management techniques that were already being used in the Grid community. Overall, S-OGSA is based on three key ideas: • The GIN leaders gathered at CERN (with some participating remotely via Access Grid) to determine goals and priorities for 2007. The S-OGSA model defines Grid-Entities, Knowledge-Entities and Semantic-Bindings. The idea seemed to be that Semantic-Bindings allow Grid-Entities to make use of particular knowledge sources (such as rules, ontologies, etc). Semantic-Bindings might have state, and were modelled as assertions on Grid-Entities. 3.31.14 ETSI Meeting – Nov 06 Dr Pickles attended the ETSI meeting held at the CoreGrid conference in Sophia-Antipolis, 30 Nov1 Dec 2006. • Dr Pickles represented the OGF at the third meeting of ETSI’s Technical Committee on GRID, which has won significant European funding. The meeting worked out where the ETSI and OGF agendas overlapped and strategies in which the two organisations could co-operate effectively in areas of mutual interest. In the following months, Dr Pickles invested considerable effort in arranging to bring ETSI to OGF20. 3.32 a. Provisioning Service The Provisioning Service enabled the management (creation, removal, update, etc) of Semantic-Bindings and KnowledgeEntities b. Semantic-Aware Service The Semantic-Aware Service extended existing services (such as the authorisation service) with semantic information. Prof Omer Rana is professor of Performance Engineering in the School of Computer Science at Cardiff University. 3.32.1 GGF16 For Prof Rana, a key message from GGF16 was the increasing focus on using Grid standards within industry, especially with the recent merger between the Enterprise Grid Alliance (EGA) and the GGF. Prof Rana participated in the workshop organised by the Semantic Grids research group, and the GRAAP working group sessions. Semantic Grids The Semantic Grids (SG) group organised a workshop focusing on the use of RDF/RDF-S as a way to extend descriptions of services within Grid middleware and applications. The workshop was primarily intended to increase awareness of SG technologies, with a particular focus on the EU OntoGrid project. The workshop was divided into three sessions: Page 138 S-OGSA Capabilities The S-OGSA capabilities would include all the semantically-enhanced services made available over the Grid. Two infrastructure services had also been identified: Prof Omer Rana – Cardiff For a summary of Prof Rana’s GridNet2 activties, see section 2.32, Prof Omer Rana – Cardiff. S-OGSA Model • S-OGSA Mechanisms The S-OGSA mechanisms related to how the model and capabilities described above could be mapped to existing Grid infrastructure. It seemed that the Common Information Model (CIM) was being widely used in S-OGSA to represent Grid infrastructure. S-OGSA was an interesting approach which attempted to “extend” rather than “re-create”. Presumably, this was undertaken to ensure that there was a good adoption of the approach within the existing Grid community. A related infrastructure talk by Jane Hunter focused on extending the metadata descriptions in the Storage Resource Broker (SRB) with RDF. This was suggested as a technique to extend the search capability that can be made available in the SRB. An extension to the MySRB interface was provided via the following: 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff 1. A portal with semantic search capability 2. A search engine that makes use of the RDF descriptions Within the portal, users are required to load ontologies, and then build instance data based on these, using extensions to existing SRB data types. Hunter claimed that no significant modification was made to the SRB core software, and therefore this approach could still be used alongside existing SRB installations. Although the benefit of the proposed approach was clear, it was not clear why this approach was specifically discussing SRB, because it had a much wider applicability. The approach also shared some similarities with the MIT Simile 3 project, which has developed a plug-in for the FireFox browser to support ontologies (called Piggy Bank) 4. Semantic Grids: Use Cases/Applications and Combining Technologies Three Grid application use cases which currently make use of semantic information were presented: • Comb-e-Chem – primarily focusing on workflow descriptions and electronic notebooks • Satellite mission systems (from OntoGrid) – primarily focusing on data integration from multiple types of instruments • Music Information Retrieval – primarily focusing on time series retrieval and digital libraries of music All three uses cases made use of RDF-like descriptions and had requirements for workflow, provenance and (large scale) data management. The Combining Technologies session involved a discussion about integrating event management for sensor-based information (primarily in the Geographical Information Systems domain). managing services made available over the Grid and by establishing a bridge between agent and Web Service standards, it was possible to use agents alongside existing Web Service implementations. Suguri did not, however, outline the development effort that would be needed to adopt such an approach. A new task force set up by the IEEE to address this concern was highlighted. Semantic Grids: Technologies and Standards The final session in the workshop focused on how OGSA-DAI and semantic services could be integrated. Two approaches were discussed: • OGSA-DAI-RDF • WS-DAIOnt (in OntoGrid) OGSA-DAI-RDF aimed to utilise the activity framework in OGSA-DAI to support RDF-based queries (defined in SPARQL). Once this had been achieved, it might also be possible to transform SPARQL (using XSLT) to other representations. WS-DAIOnt, on the other hand, was primarily focused on providing access mechanisms for ontology services which had been wrapped as WS-DAI services. Another focus was the ability to extend the data access mechanisms so that RDF-based queries (using Sesame/SeRQL) could be made to these ontology services. Overall, the workshop usefully highlighted work being undertaken in the Semantic Grid area. The use of RDF was a significant common theme running through the talks, which may have been by design. The use of SLAs was not really addressed, although it was clear that the approaches currently being used to define and store metadata in RDF could be used to provide the following: • Geoffrey Fox also presented a number of registry services that could be used to record service metadata for use in the composition of services. A number of WS-X standards were discussed with relevance to the dynamic composition of services, with a key focus on improving messaging performance between services. A second talk, by Hiroki Suguri, focused on the integration of the FIPA agent standard with Web Services. Suguri discussed the limited capability for representing semantic information within WSDL, and compared this with “agent capability” descriptions provided in FIPA, along with agent interaction support through FIPA performatives. A key point in the talk was the need for bridging agent capability and interaction mechanisms with Web Service standards. The motivation for this was the additional reasoning capability that an agentbased approach would provide. Suguri also reasoned that agents provided a more useful abstraction for For instance, in the current WS-Agreement specification, the wsag:Penalty and wsag:Reward are not very clearly specified. Relating these terms to SLOs in the agreement would provide a useful way to reason about the types of penalty or rewards that could be associated with violating or meeting a particular SLO. • 3 Relationships between different SLAs In this instance, it could be useful to assess the difference between two SLAs that have been generated by the same party at different times. This would also be useful for generating a composite/aggregated SLA from multiple independent SLAs. The requirement for aggregating SLAs had already been identified a number of times. • MIT, “Semantic Interoperability of Metadata and Information in unLike Environments (SIMILE)”. Available at: http://simile.mit.edu/. Last accessed: March 2006. Relationships between terms within a SLA (such as WS-Agreement) A means to store SLA templates in data stores that could be accessed via RDF-based query languages Such a capability would be particularly useful if a client wished to generate an SLA from a pre-defined template, for instance. 4 “Piggy Bank”. Available at: http://simile.mit.edu/piggy-bank/. Last accessed: March 2006. Page 139 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff • Query stored SLAs to look for conformance to particular business level objectives (different from SLOs) This would imply the use of a policy to define such objectives. Some of these concepts were discussed at subsequent sessions of the Semantic Grid research group and via email-based discussions. GRAAP-WG The need for an SLA had been outlined by a number of Grid application users. Some people had identified such SLAs to be static descriptions of the capabilities of resource providers, whilst others advocated a more dynamic approach which involve negotiation of SLAs. As Grid infrastructure moved towards commercial adoption, the use of SLAs would become more significant. The GRAAP-WG released the WS-Agreement specification on September 20, 2005 as a GWD-R. This specification had already undergone an extensive review process (via the GRAAP mailing list and telcon sessions), and the discussion at GGF 16 was centred on evaluating some of these comments and how the document could be improved. A total of 92 comments had been received, ranging from issues related to the syntax and grammar used to discussion points related to how WS-Agreement could be extended, with additional support for security, negotiation, etc. It was decided that the first port of call should be to finalise the current version of the document, and in future interactions to consider extensions proposed by the community. The current version of the WS-Agreement specification provided a collection of terms that could be used to specify an agreement, but did not identify any particular set of terms that identify what service properties were to be delivered under the agreement (see discussion below). The GRAAP group was also exploring how the WSRF specification would affect “Resource Properties” within WS-Agreement. This discussion was postponed to a future telcon. In the first GRAAP session, Toshiyuki Nakata (NEC) presented his work on utilizing multiple SLAs within an Internet Service Provider (ISP), and explained the requirements of to be able to combine terms in multiple SLAs to create an aggregate SLA. He also identified the need to define common terms within the SLA. Some discussion followed on what would be the best candidate for providing this set of terminology, based on the activities that were already taking place within the GGF. Most were of the opinion that the terms in Job Submission Description Language (JSDL) should be used, whereby JSDL elements could be composed into WS-Agreement Service Description Terms (SDTs). Essentially, SDTs identified the functionality that would be made available by a provider under the agreement. Alternatives to JSDL included the Common Information Model (CIM) that was primarily being developed by DMTF, although it was now also being evaluated at the GGF. It was generally agreed, however, that at the present time, a CIM schema in XML did not exist and JSDL would be a better candidate to consider. Another alternative included Key Performance Indicators (KPI) related to Web Services in particular. KPIs indicate measurable business-level Page 140 parameters/data that can be used to evaluate the performance of a particular business function 5. Once KPIs extracted from business processes have been determined, it is possible to identify useful relationships among SLAs through such indicators. When more than two parties were involved, and more than one process, the monitoring and controlling of such a relationship was vital. KPIs therefore focused on business-centric parameters, whereas parameters in JSDL and CIM are more system-centric. An associated question that needed to be considered was the types of monitoring tools that had to be used to record such parameters. Although a variety of tools existed for system-related parameters defined in JSDL, there might not be many such tools available for KPIs. An alternative approach in projects such as AKOGRIMO involved users specifying service bands, such as “gold”, “silver” or “bronze”, rather than specific parameters. A key discussion point remained - identifying how service parameters and constraints on these parameters needed to be specified. At the time, very little experience in the use of WS-Agreement within “real” business-oriented Grids existed. It was generally agreed that it would be useful to evaluate this, and to encourage greater uptake of WS-Agreement. The second GRAAP session started with two presentations of WS-Agreement use cases. The first presentation, about the EU CATNETS project from Michael Reinecke (University of Bayreuth) and Rana, focused on the use of WSAgreement for choosing service providers in a Grid market. The project used SLOs within WS-Agreement to define economic metrics such as “social utility” and price. The project was essentially considering the coexistence of two types of markets – a service market and a resource market. Hence, a particular service instance could co-exist on multiple computational or data resources, and would have a different price (depending on the performance profile of the underlying computational/data resource). WS-Agreement had been used in the project as a means to specify user requirements which were then subsequently translated to service discovery requests in the resource and service markets. The second presentation (by Wolfgang Ziegler) in the session discussed the VIOLA project which makes use of WS-Agreement to support meta-scheduling. Both of these presentations outlined the need for negotiation mechanisms within SLAs, and it was noted that the GRAAP-WG was currently gathering requirements on negotiation and would consider publishing an “informational” document in the near future. The third GRAAP session involved interactions with Franco Travostino (Nortel Networks) to identify how WS-Agreement would be used by network providers. He discussed the importance of generating an agreement between multiple carriers when setting up an end-to-end path. The IPSphere forum 6 was 5 N. Nagaratnam, A. Nadalin, M. Hondo, M. McIntosh, and P. Austel, ”Business-driven Application Security: From Modeling to Managing Secure Applications”, IBM Systems Journal, pp 847– 867, Vol.44, No.4, 2005 6 “IPSphere Forum”, Available at: http://www.ipsphereforum.org/home. Last accessed: March 2006. 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff discussed in this instance, which was focusing on open standards for delivering particular network quality and reliability to business users. Franco also identified the need for adding security to WS-Agreement, such as the capability for a two-sided authentication of the agreement. It was also discussed whether it would be useful to identify networking terms (similar to JSDL) that could be more widely shared in the WS-Agreement. At the time, there was also no provision in WSAgreement for a cancellation of the agreement. It was agreed that this was an important issue, and would be dealt within in a future GRAAP session. There was also general agreement that it was useful and important to establish a test bed for evaluating the use of WS-Agreement. Participation at the Interoperability Fests was identified as a useful way to achieve this. Conclusion Prof Rana primarily participated in the GRAAP and Semantic Grid groups at the GGF. He also attended the industry panel organised by Wolfgang Gentzsch. The panel included participants from the Japanese Grid/NAREGI project, the US TeraGrid, Intel, SIMDAT and NextGRID. Virtually all commercial participants still saw security as a significant barrier to adoption of Grid computing within their own companies. All of the commercial participants were now also considering the use of SLAs as a means to specify the types of capability that should be made available for use (within their own companies), and based on existing mechanisms that were already available (as a means to interact between sub-units within the same company). The following projects at GGF16 identified a need for SLAs: • 3.32.2 GRAAP – WS-Agreement The GRAAP-WG sessions on WS-Agreement were primarily intended to analyze comments received on the current specification (released in September 2005). The comments received on the specification are based on emails to the GRAAP mailing list and subsequent discussions via weekly telcons. Good progress was being made to provide a more robust version of the specification. It was felt, however, that some of the comments were out of scope for the current specification, such as support for negotiation, and these requirements would be considered in a future version of the document. Most comments had been taken into account, although some aspects needed to be considered in more detail, as follows: • Supporting expiration was, however, not straightforward, as it might involve resolving existing reservation for resources – which might not be under the direct control of the agreement initiator. It was felt that additional intermediate agreement states might need to be introduced to achieve this. Discussion on this continued. • 4. SIMDAT and NextGRID both identified a need to establish SLAs between participants. • Discussion of WS-Agreement with the Semantic Grid community and in particular, the role of policy descriptions and SLA specification based on RDF • Presentation at the GRAAP session on CATNETs project • Discussion with GRAAP participants on set up of a test bed and description of terms (as part of Service Description Terms) • Negotiation use case for WS-Agreement based on the EU CATNETs project The need to support aggregation of multiple agreements. In this case, it was necessary to combine constraints that were part of multiple agreements and merge these constraints into a single composite agreement. It was generally felt that more examples were needed before this aspect could be pursued further. Prof Rana’s particular contributions at GGF16 were: Session chair for Semantic Grid workshop The need to support the expiration/cancellation of an agreement. It was felt that this provision was necessary for bookkeeping purposes, and simply associating a time period over which an agreement was valid was not enough. It was necessary to provide a more explicit mechanism for cancellation of an existing agreement. 2. Federico Ruggieri (INFN, Italy) outlined the need to have SLAs as part of the Italian Regional Operation Centre for Grid computing, relating the use of SLAs with accounting and monitoring mechanisms made available within the Italian Grid. • GGF17 GGF-17 took place at the Tokyo International Forum in Japan and was co-located with the GridWorld event. The event attracted around 300 delegates (for GGF) and over 3,500 delegates (for GridWorld). Most of the presentations in the GridWorld event were in Japanese with limited simultaneous translation available. Prof Rana, however, found it useful to see a variety of industry presentations, ranging from hardware and software vendors to some specialist end users. Prof Rana attended the event from May 9 to May 13. 1. The keynote talk from Ian Foster clearly outlined the need for SLAs and policies. 3. Wolfgang Gentzsch and Robert Cohen identified SLAs as an important requirement for industry uptake of Grids (part of the “Production Grids Enterprise and Research” session). Contribution to the WS-Agreement specification document, and subsequent involvement in a telcon organised on March 8, 2006 • Guarantee terms in the agreement had not been modified, although some comments on this had been received. The comments primarily indicated that it was necessary to specify which service objectives had associated guarantee terms (not currently supported). Hence, in the current version of the specification, guarantee terms might not be associated with particular objectives defined in the agreement. In the same context, it was also necessary to clarify on the state of a guarantee Page 141 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff term, that is, whether it had been fulfilled, violated (and to what extent), etc. • It was also felt necessary to state the relationship between an agreement template and the created agreement in some more straight forward manner. Some of the comments received related to the requirement of dynamic SLAs, and this formed the basis for two sessions. In particular, there had been feedback from participants in the European OntoGrid project that support for negotiations was needed within WS-Agreement. Most of these participants advocated an agent-based approach and felt that work that had been undertaken within the multiagent systems community needed to be taken onboard. None of the OntoGrid participants were present at this event to discuss their particular viewpoints. The remainder of the session focused on discussing interoperability issues in WS-Agreement. For instance, how would interoperability constraints be defined in the context of Service Level Agreements, and in particular, what are the interoperability issues in the context of WS-Agreement? Some people argued that it was too early to consider interoperability, and the current focus should be on defining a stable specification. Others felt that the specification should account for possible interoperability issues that might arise when the specification is subsequently implemented. It was felt that to support interaction between different implementations, it was necessary to define, in a precise way, how a client from one application making use of WS-Agreement could talk to another. It was felt that experience from OGSA-DAI should be used to define mechanisms to support such interoperability. Some of these issues would necessarily relate to application specific data models, for instance, the terms within an agreement would be application-specific and interoperability would then require the two applications to share these terms. However, it was also necessary to define some generic terms that could be application-independent, and which all implementations of WS-Agreement must support. To demonstrate interoperability, it was felt that multiple implementations of the same specification were necessary to identify whether interoperability between the two is possible. It was also felt that advice from the GIN-CG should be sought for this. It was generally agreed that it would be useful to have an experience document to outline how WS-Agreement is currently being used, what additional capability particular users need and subsequently identify what type of operations were necessary to support interoperability. It was also necessary to undertake tests against the specification, which could only be done once the specification had been implemented. An agreement server could be set up to allow members of the community to test their specification for interoperability. The next step was to provide implementation of the agreement once the specification had been finalised. It was necessary to: 1. Identify who was involved in implementing the specification. 2. Make parts available for others to use. 3. Make parts of the specification open source. Page 142 In the same context, it was also necessary to identify “micro-specs” for additional domain specific terms that could be used in the agreement. GRAAP Session 2: Dynamic Agreements In the context of supporting dynamic agreements, two talks were given: • Dynamic SLAs by Prof Rana • Function-based WS-Agreement by Viktor Yarmolenko (University of Manchester). Prof Rana’s talk discussed the requirements for establishing dynamic SLAs, which included: • The need to modify an agreement that had already been established – especially if the agreement was used at a time much later than when the agreement had been defined. The requirement here related to comparing the cost of re-establishing a new agreement vs. being able to adapt an agreement that was already in place. • The need to support flexibility in the agreement if an agreement initiator was not fully aware of the operating environment when the agreement was defined. In this case, the agreement initiator might not have enough information to determine what to ask for from a provider. This was likely to be the case when an agreement initiator or provider operated with imprecise knowledge about the other party involved in the agreement. Based on these requirements, two types of dynamic agreements were defined: 1. Static Agreement In this case, it was necessary to identify Service Description Terms, Guarantee Terms, and Service Level Objectives (SLOs). Both Guarantee Terms and SLOs were to be precisely defined at agreement creation time. 2. Dynamic Agreement In this case, it was necessary to identify Service Description Terms and Guarantee Terms which were now defined as ranges or functions, and Service Level Objectives which were defined as ranges or as functions. The use of range-based or function-based agreements provided a useful basis for supporting dynamicity in the agreement. Examples from the European FP6 CATNETs project were used to demonstrate how dynamic agreements could be specified and used for developing a Grid resource and service market 7. 7 Liviu Joita, Omer Rana, Oscar Ardaiz, Pablo Chacin, Isaac Chao, Felix Freitag, Leandra Navarro, “Application Deployment using Catallactic Grid Middleware”, Third International Workshop on Middleware for Grid Computing, ACM/IFIP/USENIX International Middleware Conference, November 28– December 2, Grenoble, France, 2005. 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff The second talk focused on the ability to specify agreements as functions – where options are expressed as a set of variables. Variables could include terms such as start time and end time of a particular job. In this instance, a client would send a function-based agreement to a provider, who would evaluate the function locally and return to the client the type of resources that it had the ability to provide (at the time the agreement was to become valid). The aim here was to minimise the number of re-negotiations necessary to reach some consensus on values associated with agreement terms. Yarmolenko discussed a case study that demonstrated how the approach could be used in practice. 8 Currently, the focus within this work was on specifying guarantee terms as functions, and Yarmolenko compared the use of binary and fuzzy functions. A key advantage of this approach was that particular terms could be included in the agreement the value for which was not available at the time of agreement creation. This would therefore allow greater flexibility in the way that an agreement was defined. WS-Agreement in its current form could be used to support this function-based approach, although it was necessary to identify how the functions would be described using XML. It was agreed that a workshop would be organised a day before the next GGF in Washington to discuss the following: 1. Common application independent terms that may be used within an agreement, thereby leading to re-usable implementations 2. Implementations of the WS-Agreement specifications that are currently available Job Submission Description Language (JSDL) The session on JSDL primarily focused on extensions being proposed to the specification to enable description of parallel jobs, with particular focus on MPI. Two efforts in this area were presented: one from Imperial College (UK), based on the GridSAM project, and the other from the Japanese Grid initiative (as part of NAREGI). JSDL extensions in the NAREGI project were based on two types of job requirements: • Single MPI executables, which would make use of “worker side” JSDL • Multiple executables running on multiple systems at multiple sites In the first case, it was enough to provide a reference to a single MPI executable, whereas in the second case it was necessary to relate multiple MPI executables. Based on these job types, the following types of JSDL extensions were proposed: • 8 JSDL submission and JSDL execution Rizos Sakellariou and Viktor Yarmolenko, “On the Flexibility of WS-Agreement for Job Submission”, Third International Workshop on Middle-ware for Grid Computing, ACM/IFIP/USENIX International Middleware Conference, November 28–December 2, Grenoble, France, 2005. • JSDL abstract and JSDL concrete • Wrapper for different JSDL documents – primarily by providing extensions for MPI jobs Consequently, researchers in the NAREGI project have added ComplexJobInstance, JobInstance, and AssociateJobID. Other MPI specific extensions include MPIType, MPITasks, TasksPerHost, HelperCommand etc. For instance, MPITasks defines how many tasks run on a host. Some of these extensions are therefore aimed at capturing the command line arguments that are passed when executing an MPI job. The GridSAM project also proposed MPI-based extensions to JSDL and focused on identifying the minimal set of terms that could be used across all MPI versions. The GridSAM extensions were particularly focused on supporting job submission within the UK National Grid Service (NGS), which made use of a Globus-based submission interface. It was therefore necessary that the absolute minimum set of assumptions be made about the types of jobs being submitted. The work extended POSIX elements with additional terms defined in the GridSAM project. It was also clarified that GridSAM did not execute jobs, but primarily submitted jobs to the appropriate system that was then responsible for their execution. During discussion in this session, it became clear that vendors, such as IBM, had their own internal developments taking place with reference to JSDL, and such vendors were extending the JSDL specification internally. One example presented was the use of terms associated with the Tivoli workload scheduler, which was a product internal to IBM. One reason cited for these extensions was the lack of expressiveness available within the existing JSDL specification; it was outlined that IBM requirements were not being met with the existing specification as it was too coarse-grained for their internal use. However, IBM was clearly interested in participating in the JSDL group within GGF, and making contributions based on their use of this specification. It was also made clear that JSDL primarily provided a basis for job execution, and was not intending to provide a programming model. As various JSDL extensions were being proposed, it was recognised that using very specific terms in JSDL might be too restrictive, thereby leading to incompatibility between different extensions. It was felt, therefore, that a more open symmetric matching scheme should be employed, allowing developers to add more complex attributes if necessary. It was identified that Condor ClassAds primarily provided a set of “conventions”; however, these can be extended in arbitrary ways. In this context, the relationship between JSDL and Condor ClassAds was discussed, in addition to similarities with other projects at Boston/Harvard. These projects focused on allowing the specification of arbitrary attributes. It was therefore necessary to employ some asymmetric matching scheme that enabled a resource to advertise its capabilities using a particular set of terms. Subsequently, a job would define its own requirements using terms that may not be identical to those used for defining the capabilities of a resource. This would allow resource providers to focus on the capabilities of their own resources, and application developers to focus on their own requirements. Page 143 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff Hence, the idea that a match occurs between task requirements and resource capabilities would be incidental rather than planned. It was felt that such a match-making scheme would lead to greater use, and at the same time avoid the need for everyone to use the same set of terms. In general, it was agreed that more discussion was necessary about how JSDL descriptions could map into a more open framework. Furthermore, the resource requirements section for JSDL should act primarily as a placeholder that could be extended by developers as necessary. Hence the intention would be to not put the extensibility in the JSDL schema, and keep this outside. It was also generally agreed that the Common Information Model (CIM) provided a useful basis for adding additional terms to JSDL, although additional work was necessary to fully understand how terms in CIM could be deployed in real scenarios. The JSDL specification provided a useful basis for defining terms within a Service Level Agreement (such as WS-Agreement). Any updates being proposed to this specification therefore would increase the expressiveness of a WS-Agreement. The flexibility being proposed could have benefit for service providers and users, but would lead to greater complexity when used as part of a SLA, because it would now be more difficult to evaluate whether an agreement had been violated by a provider. Information Model The Information Model session focused on the extension of terms in JSDL to enable better match-making between a client request and a provider advertisement. A position paper by E J Stokes (from IBM) 9 formed the basis for discussion in this session. The discussion focused on the requirements for providing an information model for resources – which included the ability: • To manage resource information in the system • To advertise the capabilities of resources in the system • To express a set of requirements needed by a job that needs resources to run It was necessary to determine the granularity at which such information was to be provided. Too much detail would make the information model complex to use, and too little detail might lead to the model being unusable and lead to vendors extending the model in ways that would constrain interoperability. The focus of the discussion in this session was on the type of XML syntax (primarily name value pairs, where values could be literals or ranges) that could be used to specify the information model. For example, to advertise the capability of resource computer A, the following terms could be included: <node> <name>computerA.acme.com</name> <processor> <type>Intel</type> 9 E. J. Stokes, “Information Modeling in OGSA Position Paper”, Open Grid Services Architecture WG, May 11, 2006. Page 144 <CPUspeed>3200</CPUspeed> </processor> <OS> <type>Linux</type> <physicalMemory>3000000</physical Memory> <virtualMemory>12000000</virtual Memory> <maxProcessesPerUser>32</max ProcessesPerLimit> </OS> </node> Similarly, the requirement for some job activity 2 would be expressed as: <activity> <name>activity1</name> <processor> <type>Intel</type> <CPUCount> <=2 </CPUCount> <CPUspeed> >3000 </CPUspeed> </processor> <OS> <type>WindowsXP</type> </OS> </activity> Overall, work on the information model provided a good starting point for evaluating more useful match-making techniques. The use of ranges in XML – as advocated in this position paper – was possibly too restrictive for a real application. There had been significant work in the match-making/service discovery community on using more complex description techniques (based on RDF or OWL) that were better suited to encode such resource capability or task requirements. Clearly, a closer collaboration was needed between the OGSA-WG and the Semantic Grids-RG. The information model being developed in this WG had implications also for WS-Agreement, because one aim for the work in SLAs was to validate the provision of particular resource capability that had been defined in the SLA. The set of terms that were agreed upon in the information model and, subsequently, the mechanisms to define constraints on these terms would affect the types of applications in which WS-Agreement could be deployed. Certificate Authorities The IGTF continued to focus on developing specialist certificate authorities across the world to enable Grid users to be authenticated in some uniform manner, and enable resource sharing across multiple administrative domains. Their aim in this session was to highlight work that had already taken place within various member organisations, such as European GridPMA, work in the Asia-Pacific region, etc. The discussion focused on identifying the structure of a CA, for instance, should it be hierarchic or should it be developed as a “bridge”? Further discussion was focused on identifying what the architecture should be for this. What should the Identify Policy-OIDs that could be associated with a particular CA be? It was also felt that there was a need to verify the quality of identity tokens that had been generated by middleware. One discussion point focused on the need to build support within existing middleware to be able to support Policy-OIDs. There was a request for 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff namespace constraints and a request and decision needed on Policy-OIDs. There was a question about which middleware providers should this request be directed to. The group needed to identify the level of complexity that should be provided. There were also requirements identified for externally-defined namespace constraints so that relying parties can uniquely assign namespaces for subject identifiers to specific issuing authorities. • • Session 2: Dynamic SLAs and Negotiation • WS-Agreement use in the CATNETs project by Prof Rana • A Framework for WS-Agreement Negotiation (in OntoGrid) by Valentina Tamma • Semantic Support for WS-Agreement by David de Roure GridWorld The GridWorld had over 3,500 registered participants making this, possibly, one of the largest Grid computing events in which Prof Rana had participated. The participants were primarily from Japan, with the presentations also reflecting these particular demographics. Much of the discussion in GridWorld focused on Web Services and requirements for application deployment over distributed infrastructure. Future Actions Based on Prof Rana’s participation at GGF17, his future activities were as follows: • Provide comments on the Information Model for OGSA-WG • Contribution to dynamic WS-Agreement use case document, based on CATNETS and other SLAbased research at Cardiff • Contribution to GRAAP workshop hosted alongside GGF-18. Prof Rana also made contributions to drafting the Call for Papers for this workshop. • Use of JSDL as for match-making and discussion of extensions to JSDL. • Discussions at the GRAAP and JSDL sessions formed the basis for material in the Grid Workflow tutorial at the IEEE CCGrid 2006 conference, and discussions at the UK-Singapore collaboration session at the GridAsia 2006 conference. 3.32.3 GGF18 Prof Rana attended GGF18 event from September 9 to September 14. Prof Rana participated in the WS-Agreement workshop (for which he was also co-organiser), the GRAAP sessions, the Provenance Challenge, and the Mobile Devices and Grid session. WS-Agreement Workshop The WS-Agreement workshop was co-located with the OGF, although it was not an official OGF event. The intention was to encourage participation from attendees at the OGF and to focus discussions at the GRAAP sessions based on presentations within the workshop. The workshop schedule was as follows: • Introduction to the workshop by Wolfgang Ziegler • Session 1: Co-allocation Mechanisms • HPC4U and WS-Agreement by Matthias Hovestadt • The VIOLA project by Oliver Waeldrich AssessGrid: Risk Management in the Grid by Karim Djemame and Matthias Hovestadt The HPC4U project (www.hpc4u.eu) focused on SLA-aware resource management, supporting runtime adaptation and reliability by fault tolerant mechanisms. The key focus in the project was the development of software-only solutions for an SLAaware fault tolerant infrastructure, providing reliability and QoS (acting as active Grid components). The key features included: • Definition and implementation of SLAs • Resource reservation for providing guaranteed QoS In HPC4U a resource management system (RMS) negotiates to define the content of new SLAs. Such an RMS was expected to provide interface to Grid middleware systems, and a negotiation module had been introduced as a new RMS component. New SLAs are only allowed if they can realise current resource properties (that is, they do not violate existing constraints defined by the resource providers). In HPC4U, a cluster may be considered as an active Grid component. The RMS negotiates with the Grid infrastructure on spare resources and supports migration in case of resource failures, thereby leading to an increased level of fault tolerance. Job assignments are provided to nodes with particular performance characteristics, thereby leading to the development of a topology dependent mapping. An SLA template is generated, and a filled out template submitted. This template can either be accepted or cancelled. (Currently, there is no support for negotiation and no counter proposals are generated). In the future, the RMS would drive the negotiation process by evaluating the current situation and accepting new SLAs based on current resource usage. In this context, QoS-specific Service Description Terms include: • Storage: redundancy, min. speed, etc • Checkpoint frequency • Security policy • Deadline for job completion (best effort, in-advance reservation, etc) It was suggested that the use of JSDL would be the most appropriate representation of this, as this could also be extended with additional domain dependent terms. The HPC4U project felt that implementation of the WS-Agreement protocol stack was a time-consuming process, and that they might not be able to complete a full implementation of the specification. They indicated that they would instead be able to evaluate Page 145 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff WS-Agreement implementations from other projects, if these were made available to them, and report back on their benefits within HPC4U. They were therefore willing to work with other projects to gather new requirements and test/evaluate whether these systems can be coupled as part of interoperability testing. It was also indicated that HPC4U project would be able to provide use cases to the WSAgreement effort. The WS-Agreement for Risk Management in the Grid presentation primarily focused on work to be undertaken in the recently funded AssessGrid project. A key focus was investigating SLA violations due to various reasons (such as denial of service attacks), and identifying penalties to be associated with these violations. Such violations could also be used to identify which party had been involved in breaking the agreement. Such risk assessment could also be used to identify bottleneck indicators for system planning, such as acceptable price and penalty regarding current risk and effort. A key question to be considered was the risk associated with assigning an SLA to a provider. In this way, a provider could estimate reliability of service provision (for their own services). Risk indicators were introduced as a core part of SLA negotiation, primarily to present risk in a way that would help to improve confidence/trustfulness of a service. It was pointed out that such risk assessment was intended to help a provider improve its own service provision capability, and not intended to support building of “reputation” of a provider within a system (at least in the first instance). The VIOLA project involved the development of a meta-scheduling service to distribute parallel jobs, providing insurance that the application jobs would have suitable resources to execute. A reservation mechanism was supported to provide such guarantees. The project made use of a UNICORE client which sent requests to a meta-scheduler (using a WS-Agreement template). The meta-scheduler determined the earliest time to run a job – requiring reservation of requested resources - and returned an ID. It was necessary, therefore, to have capabilities to reserve resources in the network and bind to run time data about reservation information. It was assumed that reservations are valid for a particular time period. WS-Agreement does not provide native support for agreement negotiation, and the VIOLA project therefore proposed different types of agreements (to achieve reliable co-allocation) to address this deficiency. Three types of agreements are being defined: • Declaration of Intention Agreement These specify an “intention to provide” (and may be unreliable and short lived). These agreements may be changed by the provider to reflect current resource usage, and have no costs, rewards or penalties associated with them. • Preparation Agreement These have a restricted lifetime, but are more reliable than Declaration of Intention agreements. They represent a prior reservation of resources and may have associated costs, rewards and penalties depending on the charging policy being used. Page 146 • Commit Agreement These extend Preparation Agreements (PAs), and may be used to confirm that a resource user is going to use the previously reserved resources in a PA. Based on these, the costs, rewards and penalties become effective. It was possible that conflicts could appear at each of these three levels. There was currently no specific language or description scheme in place to specify costs, rewards and penalties. There was also an intention to support multi-party SLAs. Essentially, to begin with, a single SLA from a client was identified. This was then sent to a metascheduler, which split this into multiple agreements, aggregated results and presented these back to the user. During the discussion session, the CREMONA implementation of WS-Agreement (from IBM) was identified. It was mentioned that this has a strict IPR policy and projects wishing to use this may not be able to access the source code. The EU NEXTGrid project was also considering WS-Agreement, although they had decided to implement their own SLA model. It was also generally agreed that the Job Submission Description Language (JSDL) provided an important ‘inner language’ for specifying resource properties, and should be used within an interoperability test. The dynamic SLAs and negotiation session primarily focused on the need to extend WS-Agreement with support for negotiation. Two possible directions were identified: • Extend WS-Agreement protocol to allow a multishot interaction process. This would also imply inclusion of more complex message types. Some of these extensions had already been reported in literature 10. • Develop negotiation mechanisms as a layer above the existing WS-Agreement to enable a “compositional” approach (as already proposed in WS-Agreement, by separation into the agreement layer and the service layer). Based on this approach, any negotiation mechanism may be used, eventually resulting in some SLA that could then be specified with WSAgreement. The OntoGrid was developing mechanisms to structure service exchanges between a client and provider. The project provided well defined templates for specifying agreements, such as service description terms and guarantee terms. The aim was to use the ContractNet protocol that makes use of WSAgreement based messages. WS-Agreement with WSContract Net Protocol (WS-CNP) therefore provided a refinement of decision making process, and allowed for multiparty decision making. Other negotiation protocols were also being considered in the project, such as an extended Contract Net and English Auctions. First deployment was expected by end of September. 10 S. Paurobally and N. R. Jennings (2005). Developing Agent Web Service Agreements. The 2005 IEEE/WIC/ACM International Conference on Web Intelligence (WI 2005). Sept. 2005, France. 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff Dave De Roure identified how the Semantic Grid approaches could be useful for WS-Agreement activities, in particular how the RDF/RDFS based representation could be used to encode relationships between service users and providers. He emphasised the need to better understand work being undertaken in negotiation mechanisms within the multi-agent systems community, and to evaluate how this existing work affected WS-Agreement efforts. He provided a number of examples where negotiation mechanisms could prove useful, based on existing e-Science projects that have been reported at the GGF Semantic Grid events. performed. Furthermore, the volatility of the environment of a mobile Grid made it difficult, nigh impossible, to agree on one single fixed SLA for the duration of interaction. A differentiation was made between a BaseVO (one that was always expected to be present) and an OptVO (which was dynamically created). The negotiation was carried out in four steps: 1. Make use of a set of templates (BaseVO creation) so that everything is possible. 2. Move from templates to “bundles”. This could be online, based on semantic descriptions of QoS, as one template could be instantiated several times. GRAAP Sessions The GRAAP sessions aimed to learn from the outcomes of the workshop, and identify ways in which WS-Agreement could be used to support negotiation. The first session was primarily devoted to explaining the general structure and use of the agreement to those new to the specification. The second session started with a presentation about SLA negotiation in NextGRID and Akogrimo projects, and was delivered by Bastian Koller from HLRSStuttgart. He explained the creation of a NextGRID SLA Schema as a way to support generalised specification for SLA Management. The approach consisted of a customer domain and a service provider domain. The customer domain consisted of a Business Level Object (BLO) component that was transferred to a “Negotiator”, also maintained by a customer. Similarly, a service provider would also maintain a Negotiator which would interact with the same component in the customer domain. Interaction was achieved by sending bids to the Negotiator (from the customer) and obtaining offers (via an Offer Manager) from the provider. A service provider would make an offer based on QoS History, the BLO component involved, etc. This could be one of the following: • A single shot process – a customer may only choose from a set of Offers (and may not ask for another bid/offer) – thereby leading to a focus on a discrete offer protocol • A multi-phase negotiation, where the discrete offer protocol was extended to n-phases The next steps in the project were to implement the VO management components involved in managing bids and offers based on SLAs. It was expected that there would be a WS-Agreement implementation also in the project from the Jeulich Research Center. In the AkoGrimo project, users and providers of Grid Services can be nomadic or mobile. Such users and providers therefore need to deal with a fast changing context (bandwidth, quality of connection, device capabilities, etc). Hence, users and providers have some kind of contract that limits the potential services that can be used. In the project, SLA negotiation was found to be a useful requirement and was implemented in way that was very close to the WS-Agreement specification. It was explained that the negotiation process already took a long time, even when considering pre-defined agreement templates. In Akogrimo, both static (non-negotiable) and negotiable terms in SLA needed to be present to ensure that legal requirements were met, and to limit the set of parameters on which negotiation could be 3. Employ service discovery using the desired profile/bundle as a search parameter to filter out service providers that would never say “yes” in a negotiation. 4. Set up mechanisms to choose a particular bundle. The discussion in the remainder of the session was focused on how WS-Agreement specification could be implemented, and how interoperability testing could be undertaken on different implementations. It was generally agreed that a subset of JSDL would be used in the first instance to identify terms, and implementations across various European projects would be used as the basis for interoperability tests (a necessary requirement for the GGF/OGF standardisation process). The European projects who have expressed an interest to participate include: • HPC4U • CATNETs • VIOLA Other implementations from EGEE and IBM’s CREMONA would also be explored. It was also agreed that use cases for negotiation and dynamic agreements would be useful to obtain, and three projects agreed to provide these, namely HPC4U, OntoGrid and VIOLA. This led to the third GRAAP session which focused on how negotiation mechanisms could be used alongside WS-Agreement. It was discussed that negotiation had been taken out of WS-Agreement specification because it was deemed to be too complex. There was earlier work on trying to develop a WS-Negotiation specification, but limited progress had been made on this. Various types of negotiation were identified – hence “haggling negotiation” – when the Agreement was being made, and were considered too complex to model. However, identifying the types of haggling that could be permissible could be provided as part of the creation constraints. A more useful form of negotiation was considered to be modification negotiation, that is, how could an agreement that had already been made be modified? Here, the intention was to specify modifications necessary to an already made agreement, and mechanisms that would be needed to reference a previous instance of the agreement (using an EPR). Modification specification could relate to Service Description Terms (SDTs) or the expiration period, for instance. Page 147 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff It was also necessary to identify rendezvous points, that is, places where parameters within an agreement could be modified. This could also be made part of a negotiation protocol. A related question asked at what point in the service execution life cycle should or could the contract be negotiated to ensure that there was some continuity of the on-going process? This was particularly true for a long running process where modifications to parameters needed some coordination. The modification process was also required to specify what changes would be allowed and what would constitute a new agreement, compared to a modified version of a previous one. Here ‘time’ was a parameter in the agreement that was required to be treated in a special way. It was generally agreed that additional discussion was necessary on this topic, and further work would continue via email interactions and teleconference discussions. Provenance Challenge The Provenance Challenge workshop required participants to demonstrate how their particular system could be used to answer a pre-defined set of queries on a given workflow. A key theme in the challenge was to identify common issues that were found to be of interest in different Provenance systems and ways in which queries could be answered. A total of 17 groups participated in the challenge. A large number of groups made use of SQL or relational databases to record interactions between actors involved in a workflow. Many found the relational model to be of benefit, except when trying to compare two workflow graphs. Cardiff University (represented by Vikas Deora and Prof Rana) participated in the challenge by providing visualisation tools that could be used alongside the p-struct schema developed by Southampton University. • Two talks on healthcare services that could be offered via sensor-based devices, and integration of these into a personalised health management systems • Integrating peer-2-peer technologies (P2PS) This was primarily based on the JXTA system from Sun Microsystems, with ns-2 network simulator, for better understanding communication delays and scalability issues within a sensor network. The benefits of the P2P system were outlined, along with a discussion of the AjentJ simulator used to integrate P2PS with ns2. Prof Rana delivered a talk on the DTI-funded Healthcare@Home project, which was investigating how data from medical sensors could be acquired and analysed for trends analysis. The project focused on monitoring parameters that were most relevant in the context of diabetes management, with particular focus on the recent NHS National Services Framework, identifying care pathways for individual patients. Grid-based data mining/analysis was being used in the project to undertake trends analysis. 3.32.4 Prof Rana attended OGF19 from January 28 to February 2. Prof Rana participated in the following sessions: • The GRAAP working group, focusing on WS-Agreement specification • The Semantic Web workshop, organised by Dave de Roure and Carole Goble • The OGSA group, particularly the workflow and information model discussion. • The Quality of Service BoF, organised by Steve McGough • The workshop on Reliability and Dependability, organised by Chris Dubrowski and Geoffrey Fox • The Software Providers tracks on OMII and Grid Sphere Information Model Not much progress has taken place in the OGSA Information Model group since the last GGF meeting in Tokyo. There was still an emphasis on modelling static resource properties, and relating some of the terms being used to JSDL. The most significant update involved the development of simple description examples to illustrate how the information model could be used in practice. Mobile Devices and Grid The Aggregating Mobile Devices with Grid working group has taken on various other names in the past – including Resource and Appliance Aggregation. The focus has primarily been on mechanisms to integrate wireless devices and sensor networks with Grid infrastructure. Previous talks at sessions of this group have ranged from discussion of RFID tags as a means to acquire data to mechanisms to network home appliances and use this as a means to undertake computation. This particular session had a total of five talks: • Mobile agent techniques for managing workload within a cluster/Grid environment • Integration of sensors and mobile devices for military and naval scenarios Page 148 OGF19 GRAAP Working Group Although three sessions were scheduled for the GRAAP working group, the group were able to resolve outstanding issues in two sessions. The first session focused on public comments that had been received on the WS-Agreement specification. Wolfgang Ziegler flagged the issues and identified how these issues were being addressed in the updated specification. It was expected that the specification document would be modified to reflect these comments within the four weeks following the OGF meeting. There was also discussion about WS-Agreement interoperability, and which aspects of interoperability would be most useful to demonstrate across different WS-Agreement implementations. It was generally agreed that interoperability at the level of the protocol was too simple. A teleconference was scheduled for the individuals/groups involved in the implementation of the WS-Agreement specification. Participants from the Fraunhofer Institute, Cardiff University, Juelich Research Center and University of 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff Paderborn would be invited to participate in the teleconference. Based on feedback from the OGF area directors, it was identified that an information document (that would be reviewed) would be needed to outline the operations that would be involved in the interoperability test(s). There was also discussion about the general area of dynamic SLAs, and identifying precisely what was implied by this term. For instance, many coming to WS-Agreement sessions in the past had identified the need for including negotiation within the WS-Agreement specification. However, it was, once again, generally agreed that negotiation was too complex to support in the current version of the specification. It was proposed that perhaps the group should consider negotiation profiles, similar to the HPC Profile that had been proposed by other groups at OGF. This was found to be a useful middle option, and this aspect would be considered at a future session. Semantic Web Workshop The Semantic Web 2.0 workshop focused on two main themes: • Usability issues associated with existing e-Science software • How emerging themes such as mashups and gadgets (from Google) could be used to support dynamic user interfaces A tutorial, by Pamela Fox and Marlon Pierce, on mashups and Web 2.0 technologies identified the ease of creating a mashup using the Google API. The tutorial stressed how this approach could be used to provide quick prototypes. It was recognised that there were too many Web Services (WS-*) standards currently in existence, and it was often hard to identify how they related to each other. It was also difficult for new programmer to decide which of these to focus on. Marlon Pierce demonstrated the use of the Google maps API, and how additional data could be combined with the maps provided by Google. Geoffrey Fox also indicated that mashups were similar to workflows – as they involved integration of a number of capabilities to generate an output – which could then be displayed through a Web browser. Prof Rana disagreed with this view, for the following reasons: 1. A mashup can involve combining capabilities of a number of services (in this sense, they are somewhat similar to the outcome of a workflow process); however, the actual process involved is never fully exposed, unlike a workflow. 2. Workflows involve a combination of different control strategies that indicate how service interaction should take place. Often, this control flow is also exposed to some enactment tool. In a mashup, this is not undertaken. This workshop provided a good discussion of semantic Web technologies that could be used to develop user interfaces for scientists. The relationship between mashups and portals was not fully explored, although these seemed to be (at a first glance) complementary technologies. OGSA Workflow and Information Model Discussion The OGSA information model working group discussion focused mainly on the use of XQuery over JSDL and BES documents. The work in this group was rather sluggish at the time, although good overall progress was being made by Ellen Stokes. It was not immediately clear, however, what additional benefit the approach being proposed in this group offered above Condor ClassAds and the Redline System from Argonne. In addition, significant work in semantic match-making was being ignored by the group, especially the work undertaken in the multi-agent systems community. It was also recognised that there was a need to support workflow descriptions in the OGSA architecture. The OGSA workflow discussion was led by Andrew Grimshaw (University of Virginia). The discussion was primarily focused on whether it was necessary to extend OGSA with workflow descriptions, or whether the group should interact with other external work, of which a significant amount already existed. The need for workflow was based on the inability, at present, to identify job dependencies in OGSA. A middle ground was advocated and multiple groups have been identified to do the following tasks: 1. Identify what was already in existence. 2. Explore what extensions would be necessary to support workflow in OGSA. 3. Investigate what types of workflow would be useful to see in OGSA. Cardiff were to participate in task 1. Quality of Service BoF The QoS BoF was organised due to a number of requests at OGSA working group sessions in the past. Many people had expressed an interest in providing a more concise view of QoS. The QoS BoF consisted of three talks: • The use of QoS metrics to support replica management (Mark Morgan, University of Virginia); • The G-QoSM project—identifying relationship between QoS metrics and Service Level Agreements (Prof Rana); • The use of QoS in the GridCC project (Steve McGough, Imperial College). The subsequent discussion focused on providing a more concise definition of QoS. Many at the BoF were concerned that a more precise list of metrics that could be associated with QoS were required, and often the distinction between QoS and an SLA was not clear. Members of the networks community present at the BoF identified that QoS had well-defined semantics in the context of network applications (focusing on bandwidth, latency, packet delay and packet jitter), and that similarly, the Grid community should consider the end-to-end QoS that could aggregate quality metrics across an application, middleware and the network layers. It was agreed that to keep the task manageable, a focus on QoS metrics associated with the OGSA would be a useful starting point. Page 149 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff Service Level Agreements were one of the key topics of discussion. The workshop was funded by the EPSRC “Agentcities.UK” network—with secretarial support from NeSC. This workshop was jointly organised by Julian Padget (Bath University) and Omer Rana. Members of the Semantic Web community (such as researchers in the EU TrustCom, EU OntoGrid, OMII Knoogle, EPSRC DiscoveryNet, EU SORMA, EU CATNETS and EU BREIN projects) participated in the workshop. The event was therefore also intended to provide cross fertilisation between Semantic Grids and SLAs. Reliability and Dependability Workshop The Reliability and Dependability workshop was centred on: • Providing a more precise view on reliability issues in the context of Grid applications • Mechanisms that could be used to adapt the underlying infrastructure to support reliability The workshop consisted of a number of talks from academic researchers and infrastructure providers outlining what they considered to be reliability issues, and their particular perspectives on supporting a more adaptive infrastructure. A common theme in many of the talks was the ability to monitor running applications for “exceptional” behaviour, and using actuation mechanisms to re-schedule, re-start or terminate error-prone applications. The actuation approaches were primarily based on feedback control system ideas, whereby a change in system state from some ideal would be monitored, and the system pushed to move towards this idea (or controllable) state. Example traces from an application executed over the TeraGrid was used to demonstrate the ideas. It was interesting to see many similarities between the techniques being discussed by the authors and approaches being advocated within the Autonomic Computing community. It was generally recognised, however, that supporting suitable actuation was a difficult process in a distributed environment. Software Providers Track – OMII and GridSphere The Software Providers track was being introduced at the OGF for the first time. It was a very useful introduction to many software libraries for Grid computing. The OMII track primarily focused on the OMII projects that are now available for use (such as OGSA-DAI, GRIMOIRES, etc.) and newly funded projects that are expected to deliver over the next year. The GridSphere session focused on changes that have been made to the GridSphere Portal Development Tools, and introduced the Vines software that will provide an API that can be used alongside GridSphere. The Vines API can be used in a standalone mode, and could provide capability that could be integrated with user applications. Next Steps Prof Rana was engaged in the following subsequent to this OGF meeting: • A teleconference was organised to focus on discussing interoperability tests that need to be carried out between two (or more) WS-Agreement implementations. This telcon was intended to define the basis for an information document on WS-Agreement interoperability. • A workshop at the National e-Science Centre in Edinburgh with a particular focus on agent-based Grid computing. Page 150 • A workshop was proposed at OGF20 in Manchester (in May 2007), consisting of two 90minute sessions. The focus at this workshop was on Dynamic Service Level Agreements. This workshop was jointly organised by Wolfgang Ziegler (Fraunhofer Institute), Philipp Weider (Jeulich Research Center) and Prof Rana. 3.32.5 OGF20 Prof Rana attended OGF20 in Manchester, May 7–11, 2007. A key message from this particular meeting was the increasing focus on Web 2.0 and social networkbased technologies to support collaboration between application users. Web 2.0 technologies (such as the Google Programming API, AJAX, JSON, etc) provide ease-of- use and development in comparison with many of the existing Web Services specifications. Prof Rana participated in the following activities: • Co-organiser: Dynamic Service Level Agreements Workshop • Participant: SOKU Workshop • Participant: GRAAP working group sessions • Speaker: Workflow working group session • Participant: SAGA API session • Participant: Service Level Terms for OGSA BoF GRAAP The GRAAP working group involved three sessions, focusing on the currently released specification of WS-Agreement. The first session was dedicated to presentations from European projects in the area, such as AssessGrid, which focused on aspects of risk assessment in service provisioning. The second session focused on discussion of negotiation protocols that can be used alongside WS-Agreement. The discussion focused on two key points: 1. Whether a two phase commit protocol was a useful protocol to support 2. Whether negotiation should be part of WS-Agreement, or whether it could be something undertaken outside the specification, but could make use of terms/schema of WS-Agreement. The key concern regarding issue 1 related to whether reliable messaging could be guaranteed when a provider was making an offer to a client. It was argued that message delays could lead to a provider having to wait for approval from a client, thus 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff reducing the ability of a provider to offer resources to other clients. Relationship with such models in databases was also discussed. Issue 2 remained an open concern in the GRAAP working group, as some felt that negotiation should be an integral part of the WS-Agreement specification. Prof Rana disagreed with this, especially as it was likely to make the specification more complex. The final session of GRAAP focused on identifying interoperability experiments that needed to be undertaken once the GRAAP specification had been approved. Discussion in this session mainly focused on identifying what type of interoperability work was necessary, and what needed to be implemented to demonstrate interoperability. Currently, four research groups are involved in implementing the WS-Agreement specification, to varying extents. These include the Fraunhofer Institute, the University of Stuttgart, the Tech. University of Berlin and Cardiff University. It was generally agreed that coordination between these groups was necessary, and it would be useful to identify what interoperability tests could be undertaken between these groups. A document has been started that outlines these interoperability tests, the first version of which was discussed and started at OGF20. coordinate the execution of a workflow was also stressed, where the portal would interact with workflow enactment engines, allowing a workflow to be split across such engines. Evaluation from OGF19 After the OGF meeting, Prof Rana was engaged in the following related activities: • This teleconference was intended to define the basis for an information document on WSAgreement interoperability. Outcome: The telcon was organised and led to the first working document outlining interoperability tests needed. • The Dynamic SLAs workshop was organised to emphasise the importance of identifying and managing SLAs where: 1. SLOs can change during service execution. 2. A SLO description scheme allows for changes in SLOs during service provisioning to prevent an excessive number of violations from being raised. The workshop included a number of presentations from European projects (mainly), such as AssessGrid, BREIN, BeinGrid, Grid Job Scheduling and HPC4U. A panel session to assess the importance of dynamic SLAs with reference to business and scientific applications was organised. There were three panellists in the session. Three outcomes of the panel session were: Outcome: The workshop was successfully organised with over 25 participants from the UK and Europe. TrustCom, OntoGrid, OMII, DiscoveryNet, SORMA, CATNETS, eRep, and other projects were presented. A document outlining use cases of agent-based approaches in Grid computing was started. Currently over 10 use cases have been identified. • 2. It was useful to consider intervals when defining SLA properties to ensure that a provider was able to deal with small changes in provisioning and still remain within valid limits. Outcome: The workshop was successfully organised and included participation of Karim Djemame (from Leeds University) as a coorganiser. It was expected that a use case document would be produced to address issues 1 and 2 in particular. The Sharing Workflows workshop focused on the need for sharing workflow descriptions and subsequently mechanisms to enact workflows across different engines. A variety of views were presented, from the need to support semantic annotations on abstract workflow graphs to support for plug-ins that allow workflow enactment to take place in their “native” environment. The use of a portal to A workshop was proposed for OGF20 in Manchester (in May 2007) consisting of two 90minute sessions. The focus at this workshop was on Dynamic Service Level Agreements. This workshop was jointly organised by Wolfgang Ziegler (Fraunhofer Institute), Philipp Weider (Jeulich Research Center) and Prof Rana. 3. It would be useful to identify which communities would benefit from dynamic SLAs, and whether suitable use cases could be provided to guide the development of such SLAs. Sharing Workflows Workshop A workshop was held at the National e-Science Centre in Edinburgh with a particular focus on Agent-based Grid Computing. SLAs were to form one of the key topics of discussion. The workshop was funded by the EPSRC “Agentcities.UK” network with secretarial support from NeSC. This workshop was jointly organised by Julian Padget (Bath University) and Prof Rana. Members of the Semantic Web community (such as researchers in the EU TrustCom, EU OntoGrid, OMII Knoogle, EPSRC DiscoveryNet, EU SORMA, EU CATNETS and EU BREIN projects) participated in the workshop. The event was, therefore, also intended to provide cross- fertilisation between Semantic Grids and SLAs. Dynamic SLAs Workshop 1. It was necessary to precisely define what constituted “dynamic” SLA, as compared to “static” SLAs. A teleconference was organised to focus on discussing interoperability tests that needed to be carried out between two (or more) WS-Agreement implementations. Future Activities The following future activities were planned to take place during GridNet2: • Additional work on the use-case document outlining use of agent-based approaches in Grids. • Additional work on interoperability tests and coordination between WS-Agreement implementations. Page 151 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff Prof Rana also planned and co-organised workshops at the ACM/IEEE 2007 conference. 3.32.6 OGF21 Prof Rana attended OGF21. Web 2.0 and social network-based technologies continued to provide an important new growth area. Prof Rana participated in the following activities: • Participant: GRAAP working group sessions • Participant: Arts and Humanities Research Group (newly formed) • Participant: JSDL working group • Co-organiser: GridNet2 e-Science Workshop GRAAP The GRAAP working group involved three sessions, focusing on the currently released specification of WS-Agreement. The first session was dedicated to discussion about interoperability and conformance testing of implementations to the current specification. Two implementations of WS-Agreement were identified – one from the Fraunhofer Institute and the other from Technical University of Berlin. The aim of this first session was to discuss the current status of these implementations and when these were likely to be made available to the general Grid community. A key point of discussion in this context related to identifying what aspects of the specification needed to be tested within the interoperability experiments. For instance, if the inner language was being considered, then perhaps this would be testing a job submission system (considering JSDL being used), and perhaps not the WS-Agreement specification in particular. No particular consensus was reached, and further discussion is necessary to identify use cases for interoperability testing. A Wiki has been set up, and one scenario has so far been described. The next steps in this context involve identifying additional scenarios. The second session focused on identifying “microspecs” that could be used as inner languages within WS-Agreement. Currently, JSDL, BES and a Network microspec are being considered. The group discussed whether it was necessary to consider a subset of the existing JSDL specification for interoperability experiments, and more specifically, which tags from JSDL should be used for this purpose. The third session was focused on discussion of negotiation protocols that can be used alongside WS-Agreement. It was generally agreed that negotiation was seen as an additional capability that could be built above the existing WS-Agreement specification. Because the specification had recently been approved, there was reluctance to make any modifications to the specification in its current form, until interoperability and conformance tests of existing implementations were completed. GridNet2 e-Science Workshop This workshop was organised in collaboration with Dr Ian Taylor (Cardiff) and Dr Stephen McGough (Imperial College), and had the following objectives: Page 152 1. To highlight work that the UK e-Science community was doing at the OGF and in related standards bodies (such as W3C). 2. To support interaction between people from the UK and others interested in efforts within the UK e-Science community. 3. To encourage working across different working groups because many people in GridNet2 were active participants in OGF. The workshop involved presentations from a number of GridNet2-funded researchers. More details and presentation slides are available on the Wiki: http://wiki.cs.cf.ac.uk/twiki/bin/view/Sandbox/Ope nGridForum21 The workshop demonstrated a number of overlaps in interests across the different participants funded by GridNet2. It was generally agreed that a UK dissemination event would be useful, perhaps to take place alongside the UK e-Science AHM. Such an event could focus purely on standards activities. Attendance at the workshop was rather disappointing overall. However, the discussion towards the end of the workshop, focusing on GridNet2 sustainability, was useful and interesting. JSDL Working Group The JSDL working group sessions focused on comparing terms supported in existing Grid scheduling systems, such as GridWay, Globus, UNICORE and Genesis II. The discussion covered the terms that were likely to be important when considering job submission across these different systems. The part of the discussion looking at similar experiences when developing the DRMAA API was interesting. Arts and Humanities Research Group The Arts and Humanities Research Group began with two invited talks, focusing primarily on UK and European-funded projects in this area. The first was given by Alexander Voss (Edinburgh University) and the second (via a telcon. link) by Tobias Blanke (Kings College London). It was interesting to see the very wide range of projects being undertaken in this area, and overlap of interests with other areas of e-Science, such as the significant emphasis on intellectual property rights. Arts and Humanities provide a useful new domain that could provide new requirements for e-Science. Interestingly, Tobias Blanke outlined the need for workflow systems, data management and semantic annotations to existing data archives, as being some of the key technology enablers for arts and humanities applications. 3.32.7 OGF22 Prof Rana attended OGF22. A significant focus in this particular OGF was on application users and how they are currently making use of Grid computing technologies. Prof Rana participated in the following activities: • Participant: GRAAP working group sessions • Participant: Enterprise Adoption: Enabling the Next Generation IT Infrastructure workshop 3 GridNet2 Reports 3.32 Prof Omer Rana – Cardiff • Participant: Financial Services workshop • Participant: Cloud Computing BoF • Participant: Data Management workshop GRAAP The GRAAP working group involved three sessions – focusing on two main themes: 1. Interoperability between two WS-Agreement implementations 2. Re-negotiation protocol The interoperability session began with a presentation by Oliver Waldrich on comparing two WS-Agreement implementations, one from the Fraunhofer institute and the other from the Technical University of Berlin. The aim was to demonstrate that two versions of WS-Agreement could work together, even though they did not make use of a common code base. Each implementation also made use of the agreement states discussed in the specification, and would also test the factory service used to create a new agreement template. As WS-Agreement made use of other Web Service standards such as WS-Naming and WSRF, it was apparent during this discussion that it was actually interoperability between naming conventions that had been the key stumbling block in undertaking interoperation. A proxy was used to support the translation between the two naming schemes in the demonstration provided by Oliver. It was therefore realised that additional work was needed to identify what precisely constituted WS-Agreement interoperability, and the approach adopted by other working groups – such as GIN – would be considered. The key idea would be to develop a client-server application, where Service Level Agreements (SLAs) encoded using WS-Agreement would be sent to the server to validate their conformance to the specification. The re-negotiation session focused on discussion around a paper, A Contract Re-negotiation Protocol, that had previously been submitted by Michael Parkin, Peer Hasselmeyer, Bastian Koller and Philipp Wieder, as part of work being undertaken in the European NEXTGrid/BREIN projects. The paper discussed a re-negotiation protocol that could be utilised in environments where messages could be delayed. The authors argued that the approach adopted within the WS-Agreement specification (and subsequent discussions in the GRAAP working group) focused on maintaining strong consistency of state using transactional protocols using a two phase commit approach. The authors proposed that such an approach was wasteful of resources and could also lead to their under-utilisation. They also indicated that such an approach would be difficult of use in environments where messages could be delayed, or where strong legal constraints needed to be satisfied. They, instead, proposed a protocol that utilised loose consistency. A very simple state machine representation and the associated messages that were needed within the protocol were discussed. Overall, the proposal received good feedback during the session, although there was no general agreement if this particular protocol should be adopted within a future version of WS-Agreement. Negotiation issues remain important concerns for the GRAAP working group, although due to the diverging views on the associated protocols, no particular approach has been accepted. Enterprise Adoption, Financial Services and Data Management The Enterprise Adoption and Financial Services workshops had a significant overlap in speakers and themes. Both workshops focused on how various industrial users were making use of Grid computing in-house, ranging from the use of Condor to specialist data storage facilities. It was clear from listening to the speakers that what constituted Grid computing for many of these industries differed significantly. Some people referred to their data centre as being a Grid, whereas others considered such a definition to encompass a small/specialist cluster. John Barr from the 451 Group provided a useful survey about the particular industry sectors that were interested/using Grid computing. His discussion of an API to support the financial services industry provided a useful integration point for common services that could be made available over a Grid infrastructure. However, he also mentioned that many in the financial services community were happy to agree on common infrastructure standards, but were reluctant to identify specialist services that they made available to their in-house users. Consequently, the API for financial services that he was intending to develop did not get traction in the business world, and had to be abandoned. The data management workshop primarily focused on work from the Storage Network Industry Association (SNIA) and their work on standards such as parallel NFS (in association with the IETF) and XAM. There was also discussion about supporting data management at different levels in the Grid, from information models for knowledge management to the lower-level resource management that was the focus of SNIA. Cloud Computing Geoffrey Fox led a BoF on cloud computing, and the benefits that such a technology could bring. He identified the “cloud” as being a coarse-grained abstraction compared to the “service” abstraction. Interaction between clouds would therefore involve looser coupling than between services. The exact distinction, however, was not clear to Prof Rana because much of the work in services was already attempting to address some of the concerns that Geoffrey raised. Prior to the BoF was a keynote from IBM about utilizing the Cloud abstraction to create better resource ensembles and creating high speed interconnection networks that would link these ensembles. The BoF identified the formation of Inter- and Intra-Grids, emphasising the considerable industry benefits that could arise from linking such infrastructure together, using high speed networks. Existing focus of vendors such as Amazon in their S3 and EC2 Cloud offerings indicates industry interest. The aim of the BoF was to create a new activity at the OGF to focus on this emerging technology. This was likely to be a useful new direction for the OGF. It would be useful for the OGF community to investigate the validity of the current standards they propose in the context of such cloud computing infrastructure. Prof Rana did not believe significant differences existed at the present time, although future interoperability standards in cloud computing might prove to be different from those in Web Services. Page 153 3 GridNet2 Reports 3.33 Dr Matthew Shields – Cardiff 3.33 Dr Matthew Shields – Cardiff executing workflows should not be lost through the process of sharing. • myExperiment and social software for workflow sharing (David De Roure, University of Southampton and OMII-UK) introduced Web2.0 paradigms to workflow sharing. • WHIP (Andrew Harrison, Cardiff University) sharing workflows between desktop tools and collaborative Web portals Dr Matthew Shields is a research associate at Cardiff University. For a summary of Dr Shields’ GridNet2 activites, see section 2.33, Dr Matthew Shields – Cardiff. 3.33.1 OGF20 The comments compared approaches to Web application mashup, which was really more like an agent system. WFM-RG – Sharing Workflows The WFM-RG session on sharing workflows was the most important session both in terms of Dr Shields’ interest and the basis of the proposal under which Dr Shields received the GridNet2 funding, namely to assist restarting interest and activity in the research group. Dr Shields’ contribution to the session consisted of taking notes and participating in the group discussions, and kick-starting the group’s research document, details of which follow. The session focused on the current perceived need to be able to share workflows for scientific applications between scientists, both within and external to the groups performing the science. Unlike previous standardisation efforts for creating a single common workflow language that could be adopted by all tools and execution engines, the session’s focus was on sharing workflows between differing tools using diverse workflow language representations. Ian Taylor opened the session by contrasting the views of the OGSA workflow group, where the predominant view was that of a standard workflow language, with the views of the WFM group members, where diverse tools and languages existed. An analogy was drawn with a conventional programming paradigm and trying to force programmers to adopt a common editor, language and compiler or interpreter. There are lots of coexisting systems so the focus should be on the scientists, helping them to share workflows. Comments on Ian’s opening included: asking if a study has been done on whether there was a real need for scientific workflow; how many Grid applications actually used workflow; and the point was made that workflows often work better with Grids than other applications so problems were often thought of in workflow terms. It was proposed that the next group output should be an OGF research report on workflow sharing. Following Ian, there were a number of presenters focusing on: • Levels of interoperability (Adrian Toth, University of Miskolc) and standardisation efforts within industry, WFMC, OASIS, OMG The comments discussed the relationship between abstract definitions (XPDL) and concrete implementations (BPEL). • Sharing workflows (Omer Rana, Cardiff University), incentives, abstract vs. concrete workflow sharing, annotations and provenance The comments included a question about how does provenance help workflow interoperability. • Workflow optimisation and sharing (Rizos Sakellariou, University of Manchester) optimisation information discovered when Page 154 • Shibboleth and security in workflow sharing approaches (Richard Sinnott, NeSC, University of Glasgow) using shibboleth attributes in creating more flexible and dynamic authorisation policies across Grids and portals for sharing workflow One of the comments was a question regarding whether this approach allowed delegation mechanisms, with the response that the service had the optional ability to allow an authorised user to further delegate. • Quality of service guarantees (Dimosthenis Kyriazis, Telecommunications Laboratory, National Technical University of Athens), an approach to mapping abstract workflow and QoS parameters to concrete federated providers. The comments included a question about how the SLA was monitored and the answer was that there was a separate component used by all workflow engines. After the presentations, the meeting content was summarised and a request for contributions to the research document made. GridNet2 Advisory Board As this was Dr Shields’ first GridNet2 award, he attended the GNAB2 meeting to observe the workings of the grant and get a wider appreciation of the sorts of activities being undertaken under it. Other Sessions Dr Shields participated in a number of other OGF sessions including some of EGEE Workflow Convener meetings, the EU presentation, and one of the Astronomical Virtual Observatory meetings. Summary Although this was not Dr Shields’ first OGF/GGF meeting, it was the first funded under GridNet2. As such, he concentrated on the workflow aspects of the meeting which met the goals of his original proposal. The main reason for attending was the Sharing Workflows session held by the WFM- RG which proved to be a worthwhile and interesting session. The approach to workflow sharing among differing tools and languages rather than attempting to create a “one size fits all” standardisation effort seemed to strike a chord with the meeting participants. 3 GridNet2 Reports 3.33 Dr Matthew Shields – Cardiff Actions • Ensure the WFM-RG Web pages are updated with the meeting summary and presentation slides. • Kick-start the WFM-RG research document on workflow sharing and solicit contributions before OGF21 in Seattle 3.33.2 OGF21 WFM-RG – Workflow Sharing and Interoperability The WFM-RG session on workflow sharing and interoperability was the key session Dr Shields attended as it was the basis of the proposal under which he received GridNet2 funding, namely to assist re-starting interest and activity in the research group. Dr Shields was the first presenter at the session, after Ian Taylor’s introduction, giving an overview of the previous session from OGF20 in Manchester and summarising the activity on developing workflow sharing and interoperability use cases in the period between the meetings. As part of his duties as co-secretary for the group, Dr Shields also took the meeting minutes and agreed to continue the organisation of the use case collection document. The other presentations included: • Ewa Deelman (ISI) summarising a recent NSF/Mellon sponsored workshop entitled A Workshop on Scientific and Scholarly Workflow Cyberinfrastructure: Improving Interoperability, Sustainability and Platform Convergence in Scientific and Scholarly Workflow. The findings of this workshop were of particular interest to the group since they outlined key challenges for the workflow community including fault tolerance, parallelism and long running workflows, and highlighted the fact that scientists don’t want a “one-size-fits-all” solution but they do want to be able to reuse abstract workflow descriptions across systems, and they do want an easier way to discover a given workflow tool’s capabilities to enable the comparison of the available tools. • Andrew Harrison (Cardiff), with his perspective on workflow embedding, which is the ability to run workflows from one tool within the workflow of another. He focused on the ability to share data and discussed RESTful approaches to how that might be achieved; • Maurizo Melato (NICE srl) gave an overview of A-WARE, a Web-based Grid workflow system. Melato talked about making use of standards such as BPEL, JBI (ESB), Business Process Modelling Notation (BPMN), and a workflow lifecycle consisting of design (method user), grounding (IT user) service binding and data mapping, deployment (IT user), submission (end user). • After the presenters had finished, the meeting moved onto a round table discussion, with the group tabling a research document for workflow sharing and interoperability use cases. Topics included interoperability between Triana/Pegasus, the scope and representation of workflow languages, outreach to the WFMC standards body, and the need for a workflow comparison document. After the discussion the meeting content was summarised and a renewed request for contributions to the use case research document was made via the group mailing list. GridNet2 e-Science Workshop Dr Shields attended the GridNet2 e-Science Workshop and presented a summary of the work being undertaken at Cardiff that GridNet2 funding has helped. The presentation included a summary of the key points from the earlier WFM-RG workshops at both OGF 20 and 21 and outlined Cardiff’s view on workflow sharing. Key to this was to focus on finding use cases, and in particular, to focus on sharing data, making sending and receiving it as simple as possible, a common theme in Dr Shields’ activities at this OGF. During the discussions concerning the future of GridNet, Dr Shields expressed his willingness to support a new application for funding and re-confirmed his intent to continue collaborating with the colleagues he has interacted with as part of the current funding. Other Sessions Dr Shields participated in a number of other OGF sessions, including some of the Web 2.0 sessions, myExperiment workshops and OGSA workflow meetings. The myExperiment workshops were of particular interest and relevance to him since some of the current work on Triana is to do with myExperiment integration, workflow sharing and interoperability. The OGF meeting was a good opportunity to hear what the myExperiment developers had to say, especially in light of other interested parties in some well-attended sessions. Summary This second OGF funded through GridNet2 helped cement some of the developments made in the area of workflow sharing from OGF20 and the discussions on interoperability in this OGF. Dr Shields’ continued focus at OGF meetings was with workflow subjects in general and the WFM-RG, in particular, and as such, the money provided by GridNet2 to enable him to attend both of these meetings has been invaluable. Dr Shields was keen to continue with GridNet funding as he thought it gave a unique opportunity for the UK e-Science community, especially newer researchers, to interact with peers nationally, and more importantly, internationally. He would support any new GridNet application and had expressed his commitment to helping any such effort. Actions • Ensure the WFM-RG Web pages are updated with the meeting summary and presentation slides. • Organise the WFM-RG research document on workflow sharing and interoperability. David de Roure (Southampton) gave an overview of myExperiment and discussed distributed services, scientists and the “social life of workflows.” Page 155 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC 3.34 Prof Richard Sinnott – NeSC Prof Richard Sinnott is the Technical Director of the National e-Science Centre at the University of Glasgow. Prof Sinnott’s GridNet2 award also covered the following people: • Oluwafemi Ajayi For more details, see section 3.1, Mr Oluwafemi Ajayi – NeSC. • Jipu Jiang For more details, see section3.19, Mr Jipu Jiang – NeSC. • Anthony Stell For more details, see section 3.36, Mr Anthony Stell – NeSC. For summaries of the GridNet2 activities covered by this award, see the following sections: • Section 2.1, Mr Oluwafemi Ajayi – NeSC • Section 2.19, Mr Jipu Jiang – NeSC • Section 2.34, Prof Richard Sinnott - NeSC • Section 2.36, Mr Anthony Stell – NeSC 3.34.1 GGF15 Prof Sinnott attended the following sessions at GGF15: • GGF15 Opening – Mark Linesch • e-Science and Cyberinfrastructure – The Middleware Challenge, Tony Hey, Vice President, Technical Computing for Microsoft Corporation • OGSA-Authz-WG • Community Activity: Leveraging Site Infrastructure for Multi-Site Grids • Business Case for Why Leading IT Organizations are Adopting Grids • Entertainment and Digital Content • What is the Software Licensing Model for Grids? Relevance to NeSC The security focus of NeSC-Glasgow (including AAA action line in UK ETF and numerous other projects, such as BRIDGES, VOTES, DyVOSE, GHI, GEMEPS, ESP-Grid, GEODE) provided Prof Sinnott’s main motivation for attending the GGF15 meeting. In particular, the session on Community Activity: Leveraging Site Infrastructure for Multi-Site Grids had numerous highly relevant presentations and was the primary purpose for Prof Sinnott’s attending the GGF meeting. The workshop itself explored how Grid security technologies could be used for VOs spanning multiple sites and how existing security infrastructures could Page 156 be leveraged. A key focus of the workshop was Shibboleth and advanced authorisation infrastructures, which was highly relevant to NeSC Glasgow projects. Ken Klingenstein gave a presentation on the Internet2 Shibboleth middleware. The uptake and widespread acceptance of Shibboleth were described, along with the wider issues of policy and legal frameworks. An outline of the challenges that had been faced in the US in terms of privacy protection and attribute release via Shibboleth were described. This was something that had not been touched yet by the Grid community but might well be something that becomes more important. Arnie Miles gave an interesting presentation showing how they had integrated Condor and Shibboleth at Georgetown University. This might well be something Prof Sinnott’s team explore at NeSC Glasgow in their Shibboleth evaluation and early adoption plans. Sinnott would like to get one of his supervision students to look into this in more detail. Jim Basney gave his MyProxy talk and outlined how this middleware could be used for unified site authentication. Prof Sinnott’s team used this software already in several projects, so recent developments were well worth monitoring. Prof Sinnott did not hear anything in this talk that was new to him. Marty Humphrey gave a talk outlining how his group had integrated MyProxy with Pubcookie. This was a fairly complex talk and not really relevant to NeSC Glasgow’s activities. The idea of Pubcookie and how they had integrated it with MyProxy did not align with the way in which Prof Sinnott’s team had been using PubCookie within the Shibboleth explorations at NeSC Glasgow. Von Welch gave his GridShib talk. Prof Sinnott was already aware of the designs and plans of GridShib that Von Welch and his team had been following, and so there were no real surprises. That said, he did show an initial implementation of this middleware in action. Prof Sinnott’s team provided an evaluation of this software as part of their ESP-Grid project with Oxford University. David Chadwick gave a presentation and demonstration of the dynamic delegation of authority between sites software put together within the DyVOSE project (led by NeSC Glasgow). This was well received. Tom Barton also gave an interesting talk and demonstration of the Signet and Grouper software for distributed attribute administration. This looked highly relevant to NeSC Glasgow activities and offered similar functionality to the existing PERMIS software but was purely web based. Prof Sinnott planned to have a final year undergraduate or postgraduate student to explore this software, which was then at beta level. The panel discussion at the end of this workshop was quite informative. It was clear that the security folk were following an aggressive technology push approach, but needed real live applications to ensure that this met the needs of the wider Grid community. This was where NeSC Glasgow fitted in. It was important that real needs and real standards were combined, and hence the work on most NeSC Glasgow projects was exploring latest developments in this area. 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC Prof Sinnott noted that he would have liked to attend the Privacy and Security in Health Grids workshop also, but this ran in parallel with the security workshop, which was a clash that should have been avoided. The other meetings Prof Sinnott attended were only indirectly relevant to NeSC Glasgow activities. The licensing meeting was interesting, but did not come up with real solutions or concrete proposals for how licensing issues and Grid technologies could be integrated. Similarly, there were no concrete ideas or suggestions for standards in this area. The entertainment and digital content workshop was of interest based upon the PGPGrid work. However, once again, this was not offering anything new; it just looked at how Grids were being used by major animation companies for rendering etc. One interesting discussion did take place on security. Prof Sinnott had not previously considered splitting the data and jobs so that compromise of data/jobs on one site would not result in the movie frames being accessed and so on, as you need all jobs and data composed to get the video frames worth stealing. There might well be scenarios in other NeSC Glasgow projects where such issues come about, for example, statistical disclosure in clinical trials. 3.34.2 GGF16 Prof Sinnott attended the following meetings at GGF16: • GGF16 Keynote – Ian Foster • Update from UK/Italy - Neil Geddes • Semantic Grid 101 • Grid Education and Training • ShibGrid workshop #1 and #2 • caGrid • Security Area Meeting • CAOPS • Town Meeting • LSG-RG Relevance to NeSC The security and life science focus of NeSC-Glasgow (including projects such as VOTES, DyVOSE, GHI, GEMEPS, ESP-Grid, GEODE, and GLASS) provided the main motivation for attending the GGF16 meeting. Specifically, the session on ShibGrid and the life science research group was the primary purpose for Prof Sinnott’s attending GGF16. Prof Sinnott gave a presentation at the ShibGrid meeting demonstrating how they had already successfully demonstrated the integration of Shibboleth and Grid technologies at NeSC in Glasgow. It appeared that they were ahead of the rest in this regard, who seemed to be still working on architectural aspects and interaction scenarios to support Shib and Grid integration (or Grid and Shib integration). The role of VOs and having multiple federated attribute repositories was discussed. There were different possibilities here which needed to be worked through. The group decided to push forward on this and to share ideas by establishing an email list and setting up a test infrastructure for exploring Shib and Grid integration. Nate Klingenstein gave an update on SAML 2.0 and Shibboleth 2.0 and 2.1 developments which would change some of the underlying technologies, and as usual, the claim was that this would be for the better. It was also claimed that these would be backwardcompatible with Shibboleth 1.3, and hence existing solutions should not be adversely affected. Time would tell how the landscape and the wider Shib and Grid efforts would be affected. David Chadwick gave his PERMIS+Grid+Shib talk on which there was not much comment. Prof Sinnott’s team have been exploring many of these things directly in the DyVOSE project. Von Welch gave a talk on ShibGrid and what the Globus folk want to do. This was a repetition of previous talks and they were still working back-to-front in that they access the Grid in the usual way and then use Shib to return attributes, which was basically wrong in Prof Sinnott’s opinion. Erik Vullings gave an interesting talk on the Australian MAMS project which seemed to be the most advanced of the rest of the groups. They had implemented attribute release tools which might well be worth exploring in future. Prof Sinnott suggested that they might do some joined-up efforts with him and his group at Macquarie University in Sydney. The UK GridShib and SHEBANGS projects were presented, but these had not started yet. Oxford (GridShib) was looking at starting with a requirements gathering process, which concerns Prof Sinnott as it was only a 1 year project. The SHEBANGS project had concrete ideas for how things would pan out but their designs seemed unduly complicated, a fact noted by Nate Klingstein who identified that there were many points of failure. That said, they had started with detailed ideas of how things would pan out where there were multiple identity providers and attribute authorities, and how this would work with Grid-based scenarios using MyProxy. The Semantic Grid workshop was interesting and might be of relevance in future. It was extremely well attended with 100+ folk crammed in. There were a couple of things that were not 100% clear to Prof Sinnott on how ontologies and data standards would actually work. It seemed that one only has to focus on OWL now as this incorporates DAML+OIL and RDF. The semantic Grid folk were looking at using semantic Grid for everything from user descriptions to resource broking and security infrastructures. It was worth monitoring in future (and possibly getting a couple of students to do some explorations with this stuff for the future). The caGrid meeting was of relevance to VOTES and GHI. They were a huge healthcare project in the US (over $100M and over 500 partners) looking at cancer-related research and healthcare in general. They had already been in contact with Prof Sinnott regarding synergies with VOTES efforts. They described a very detailed methodology but when asked, they still did not have any real (patient) data. It was worth looking at what they have been up to though to see if Prof Sinnott’s team could learn anything. They had a security report for healthcare which was worth looking into. Prof Sinnott was not sure why they had a full session just for themselves; it Page 157 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC would have been more useful if Prof Sinnott’s team had been invited to compare their solutions also. The security area meeting was a recap on what the groups had been looking into, plus a presentation from Microsoft who had plans in this area. They described the numerous technologies which they wanted to use to support delegation of authorisation and enterprise security integration more generally. This was interesting, but when asked, they did not comment on how they would or would not be interoperating with existing GGF efforts. It was clear that they have plans in the Grid arena and security was very much part of these plans. 3.34.3 GGF18 Prof Sinnott attended the following sessions at GGF18: • GGF18 Opening (Mark Linesch) • Shib+Grid workshop #1 and #2 (Von Welch) • Dynamic Negotiation • Town Hall • Security Area Meeting (Blair Dillaway) The Grid Education workshop was of relevance due to the teaching Prof Sinnott’s team already undertake at NeSC Glasgow. Prof Sinnott was happy with the idea of sharing information on how best to train and educate and would ideally like to be more involved in these activities. He was not sure what they needed to get from others as there is a successful full course established at NeSC Glasgow, but was willing to help others. • Business Grids Keynote Panel Discussion (numerous) • Electronics Design Grid Requirements (numerous) • Pharmaceutical Grid Requirements (numerous) The town meeting was not especially interesting, and the CAOPS meeting seemed to be looking for a home for the group within GGF, and the meeting was also not especially enlightening. However, Prof Sinnott only caught the latter half of this meeting, due to discussions with other attendees during the break. The security and life science focus of NeSC-Glasgow (including numerous projects such as VOTES, DyVOSE, GHI, GEMEPS, ESP-Grid, GEODE, GLASS, and OGC Collision) continued to provide the main motivation for attending the GGF18 meeting. Specifically, the session on Shib+Grid was Prof Sinnott’s primary reason for attending. The Electronics Design and the Pharmaceutical Grid requirements meeting were also highly relevant, as well, with regard to upcoming projects such as nanoCMOS (for electronics design) and VOTES, GSSFHS and potentially eCHIRURG, PERISCOPE for Pharmaceutical Grid requirements. The LSG-RG meeting was very disappointing for Prof Sinnott. There was no agenda, the organisers did not show up, and the dial-in did not work. There were around 30 people who turned up. Piotr Bala tried his best to organise the meeting and fill in for missing organisers, but it was obvious that he had no chance to really do this well. Lots of security folk attended this meeting who were probably taken in by the workshop title, but there were few concrete things discussed. They were looking for someone to come in and help drive things forward. Prof Sinnott was likely to do this upon his return. He proposed the idea of setting up a real healthcare Grid and life science Grid development activity, thereby implementing real Grids for real test purposes. This should form the basis for concrete evaluations of GGF standards and technologies, as all research groups should be doing, instead of being arbitrary talking shops. There were existing solutions which could form the basis for these efforts, such as dummy healthcare data sets, existing repositories, and existing bioinformatics Grid services and databases and so forth. Sinnott was convinced that GGF attendees and others would be interested in doing something if they thought it was for a real reason. Prof Sinnott had written a couple of documents, previously, as requested at an earlier GGF by the group’s organisers. Prof Sinnott sent these documents to the list but never had any response/feedback. This had to change or this group would fade away which would be a shame as it is very much in Prof Sinnott’s area, and lots of other people are interested, too. All-in-all, this was not a particularly fulfilling GGF for Prof Sinnott although the GridShib session was of interest. Strong leadership is needed to drive work forward and make real plans with real milestones for GGF to be worthwhile. Page 158 Relevance to NeSC Prof Sinnott gave a presentation at the Shib+Grid meeting demonstrating how his team had already successfully integrated Shibboleth and Grid technologies at NeSC in Glasgow. This included latest developments from the DyVOSE project, with dynamic delegation of trust and dynamic VO establishment, and attributes being dynamically assigned potentially to remote users, which can subsequently be used to make local authorisation decisions. Prof Sinnott gave a couple of live demos of this to people throughout the workshop. There were several interesting talks in the workshop. It was recognised by the group that there would likely be a core set of commonly accepted attributes and some others needed for Grids. This model is aligned with the work Prof Sinnott’s team have been doing in this area at Glasgow and with the UK federation efforts more generally, such as building on the eduPerson object class. Nate Klingstein gave an update on Shibboleth 2.0 and what this would look like under the hood. The NeSC Glasgow team were applications-oriented end users whereas this talk was pitched at back-end Shibboleth developers. Rather worrying was the radical change from the existing Shibboleth 1.2/1.3 approach, including complete changes to the Identity Provider, Service Provider and WAYF, and also to their protocols/interaction flows and bindings, encompassing pretty much everything as it was currently known). Prof Sinnott found these kinds of things worrying, especially with regard to formulating a strategy for UK Shibboleth efforts. 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC David Chadwick gave a very interesting talk on the latest PERMIS tools. The new policy editing tools, in particular, had several neat features, like reading in the different services from a Globus installation, web services or various others, which can then be fed into the policy definitions. • OGSA-DAI User Forum (Prof Sinnott) • VPman (David Chadwick) Similarly Erik Vullings gave an interesting update on the Australian MAMS project and took the group through the various tools the MAMS team had been building. This talk was given by skype (he was in Australia). There were numerous problems with audio throughout the Shib and Grid session unfortunately! Relevance to NeSC The Microsoft security presentation described the numerous technologies which Microsoft planned to use to support delegation of authorisation and enterprise security integration more generally. All these technologies were Windows-based, of course. Prof Sinnott was not sure how aligned this was with OGF standards development. Prof Sinnott had hoped that the Electronics Design and Pharmaceutical Grid requirements groups would have had a bit more to say. The two sessions were organised as panels and they were at a fairly highlevel. That said, the ideas covered in the Electronics Design meeting were an almost exact copy of what Prof Sinnott’s team were planning in their upcoming nanoCMOS proposal (license management, better job scheduling, ways to find/index data etc). The Pharmaceutical Grid meeting was a little generic with fairly obvious requirements which would apply to any Grid. Interestingly, Prof Sinnott thought that there would have been much more emphasis on security and building of data Grids; however, numerous panel members seemed to have HPC requirements. All-in-all, Prof Sinnott found this to be a pretty good OGF meeting, although he had to leave a day early. He would have liked to attend the OGSA-DAI User Group meeting as incumbent chair, but he did manage to speak to the OGSA-DAI team, however, and discuss aspects of my upcoming role. He would also have liked to attend the LSG-RG meeting to inform others on what NeSC Glasgow were doing in that space and try to push the efforts of the group forward. Sinnott felt that this effort was still lacking. Prof Sinnott was to be chair of the upcoming Life Science Grid conference in Japan, and the next one was to be held in Glasgow in 2007. He planned to push this forward within OGF’s LSG-RG. 3.34.4 OGF20 Prof Sinnott attended the following sessions at OGF20: • OGF20 Keynote (Peter Coveney) • BoF - Software Licensing for Grids (David Wallom/Laura McGinnis) • OGF/EGEE Plenary (Ian Bird) • Sharing Workflows (Ian Taylor/Ewa Deelman) • SAGA Overview (Thielo Kielmann/Andre Merzky) • Unified Grid Logging and Security Auditing (Martin Swany) • GridNet2 meeting (Malcolm Atkinson) Prof Sinnott also attended various exhibits throughout OGF20. Due to other commitments, Prof Sinnott was only able to attend OGF20 from Tuesday-Thursday. He would have liked also to attend meetings on Monday including the OGSA Authorisation meeting and the LSG-RG, but the timetabling of both of these (9am on Monday) meant that many would-be attendees simply could not attend. Prof Sinnott also tried to organise a workshop on e-Health at OGF20, but this was not accepted. At OGF20, Prof Sinnott gave a talk at the Sharing Workflows session organised by Ian Taylor/Ewa Deelman. It was clear that security considerations for both defining and enacting workflows were something that this community had not yet satisfactorily addressed. It was also something that NeSC Glasgow was explicitly focusing on in the EPSRC pilot project nanoCMOS. Hence, Prof Sinnott was hopeful that they would be able to push the agenda of the workflow community, in due course, with regard to security and authorisation considerations. The BoF on Software Licensing for the Grid was also highly relevant to the nanoCMOS project. Prof Sinnott was keen that this work should be pursued further, and he recommended that it should attempt to provide practical implementations showing how different license models can be supported on the Grid. NeSC Glasgow had already implemented one such scenario in nanoCMOS, integrating remote Grid services with local FlexLM installations, for example. Prof Sinnott organised the OGSA-DAI user group forum and presented his experiences of applying this technology within projects such as BRIDGES. There were several presentations given at this workshop by NeSC Glasgow personnel, covering projects such as VOTES, SBRN, GEMEPS and nanoCMOS. NeSC Glasgow planned to explore the upcoming release of OGSA-DAI since it overlapped considerably with requirements from their projects. Prof Sinnott was interested in the SAGA meeting and was keen to explore this further. The simplifications offered to Grid developers through SAGA were compelling. Given the security focus of NeSC Glasgow, Prof Sinnott attended the Unified Logging and Auditing meeting. This was an area where NeSC Glasgow did not have so much detailed experience and he was keen to ensure that they learnt from other work in this area. Prof Sinnott also attended a VPman project meeting whilst at OGF20. All-in-all, Prof Sinnott found this to be a good OGF. Page 159 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC 3.34.5 OGF21 Prof Sinnott attended the following meetings at OGF21: • Web 2.0 Social Networking for MSI Researchers • OGSA-Authz WG Meeting • Security Area Meeting • DAIS WG • OGSA-DAI User Forum • Standards All Hands Meeting • caGrid • OGSA Data Architecture • OMII-Europe • LSG-RG • GridNet2 meeting • Grid Interoperability Panel • OMII-UK Prof Sinnott left the OGF meeting on Friday morning. Relevance to NeSC All of the meetings Prof Sinnott attended had direct relevance to projects at NeSC Glasgow, with the exception of the web 2.0 workshop. The Web 2.0 workshop was the most well attended of all of the meetings in which Prof Sinnott participated. Given that this is very much orthogonal to the Grid standards effort, Prof Sinnott found this rather worrying, questioning whether or not these meetings should be taking place at OGFs or if there should be an alignment of the mashup-based approaches and Grid standards instead? It seemed that people were frustrated at the lack of stable middleware and standards and were looking elsewhere. The OGSA-Authz-WG was interesting and often had heated debates on the work and the progress of the work as a whole. It was clear that there were insufficient efforts being contributed to this WG at the time, especially with regard to the OGF standards themselves. One of the key issues was that much of this work was low-level and demanded detailed knowledge of standards and profile efforts, such as SAML and XACML, to which few can contribute. One suggestion was that this group became a Research Group, a suggestion with which Prof Sinnott agreed, since NeSC Glasgow was very much a consumer of the standards and the implementations coming from the OGS-Authz-WG. The focus should be on showing how to build secure authorisation-based systems and not just looking at developing associated standards. The security area meeting consisted largely of a presentation from Mike Jones from Microsoft on their InfoCard work. This is complementary to the security work at Glasgow using Shibboleth, LDAP-based attribute authorities and, more recently, VOMS. It was possible that NeSC Glasgow might undertake some prototyping with this InfoCard technology, in the form of a student dissertation, for example. Page 160 Prof Sinnott organised the OGSA-DAI user group forum where several presentations were given from the high-level OGF data architecture perspective (by Allan Luniewski of IBM), from the standards such as DAIS/ByteIO and their implementers at OGSA-DAI (by Ally Hume of EPCC), and from the perspective of the end user application developers with their reported experience in applying OGSA-DAI v3 in the clinical trials domain (by Anthony Stell, NeSC Glasgow). This was not the most well attended of workshops, though, which was a common occurrence during much of this OGF meeting. The caGrid presentation was interesting since it overlapped greatly with the ongoing work in NeSC projects such as VOTES. Various presentations were made on caGrid, with the focus especially on the systems being built and associated security aspects. The work was well advanced with much done in terms of ontologies and linking with domain specific knowledge. Prof Sinnott noted, however, that in reponse to his question, this system was not being used in the field by clinical trials researchers in the cancer domain. The response was that case studies/trials were coming in the future. From experiences in VOTES, Prof Sinnott was acutely aware that any solution in this domain had to be lightweight from the perspective of the clinical service provider and the end user, as they do not want to know about Grid or Shibboleth and so on. Prof Sinnott did not believe that this was the case at the time with caGrid, in that it appeared to be very much a Grid-heavy solution. The OMII-Europe talks were interesting since they addressed issues that NeSC Glasgow was facing with regards to the multitude of different Grid middleware now existing and their interoperability. Time would tell how much the interoperability scenarios that OMII-Europe were exploring and supporting would be applicable outside of their test infrastructures. The most interesting workshop at OGF for Prof Sinnott was the LSG-RG, although he could only attend the first part of this due to his presentation at the GridNet2 workshop. The workshop had a single presentation from Dr John Boyle of the Systems Biology Institute in Seattle who gave an excellent talk on why they did not use any Grid middleware at the time, even though they had tried many of them (including Globus, caGrid, Taverna, Mobius, and so on). In short, the fundamental problem Boyle identified was that this middleware comes from the top-down perspective, instead of reflecting the dynamic bottom-up approach needed by the life science researchers. The science is changing so rapidly that they need approaches that reflect the rapidly changing and evolving data, the new algorithms, and so on, and not an infrastructure per se. This resonates with NeSC Glasgow’s own experiences in projects such as BRIDGES and GEMEPS. Prof Sinnott gave talks at both the GridNet2 workshop and at the OMII-UK workshops. The GridNet2 talk was focused upon what he had done at GGF/OGF and his ideas for how to take this forward, for example, with GridNet3. For Prof Sinnott, a key aspect was the identification of OGF champions who could provide the transfer of knowledge from WG/RG groups to the wider UK community. There were enclaves of expertise in OGF which did not transfer outside of the groups, for example, people working in security didn’t know much about what was happening in workflow or data working groups, and so on. 3 GridNet2 Reports 3.34 Prof Richard Sinnott – NeSC The Grid Interoperability panel was interesting since pretty much all the middleware providers, from gLite, Globus, SRB/iRODS and CROWN through to OMII-Europe, sat on the panel. The topics discussed were broad, and in Prof Sinnott’s opinion, worrying. As a single example, the OGSA-BES specifications were highlighted as being deficient in several ways, including issues with data staging. Prof Sinnott was not sure why these issues were not identified at the outset and incorporated in to the specification without future extensions or profiles and so on. There were similar issues with other areas, such as security and information services, discussed at the session. At the OMII-UK workshop, Prof Sinnott gave a talk on the use of this middleware in the nanoCMOS project. This was well received as far as he could tell, and several people had requested his slides since then. All-in-all, Prof Sinnott found this to be a pretty good OGF. Prof Sinnott noted that this was not the best attended OGF he has been to, and wondered why his proposal to host the International Conference on Life Science Grids 2007 at this OGF meeting had not been accepted. 3.34.6 OGF22 Prof Sinnott attended the following sessions at OGF22: • Opening Session • OGSA-RUS specification • OGSA Authz • Security Area Meeting • Authz interoperation demos • OGC-OGF Collaboration (three sessions) • Encyclopaedia of Life • Financial services workshop • Keynote on Cloud Computing • Pharma, Biotech and Life Sciences workshop (two sessions) • Town Hall meeting • Grid Usage and Productivity • Data Management workshop (two sessions) Prof Sinnott left OGF22 on the Thursday afternoon. Relevance to NeSC All the meetings Prof Sinnott attended had direct relevance to projects at NeSC Glasgow. The OGSA-RUS workshop was useful in that it gave Prof Sinnott an overview of the various specifications and approaches ongoing in this area, which was especially relevant given the full economic costing and need for accounting on Glasgow resources such as ScotGrid. At the start of the meeting, the chair (Morris Riedel) actually took the time to go around the room to find out the background of attendees on the RUS specs and made sure that he covered the background information needed to bring people up to speed. Prof Sinnott thought that this should be part of many, if not all, working group meetings The OGSA Authz WG was split into two meetings. The first of these covered the discussions on the recent standards that had been put together. The VPman project, in which NeSC was directly involved, had included work on the implementation and exploitation of these standards. Prof Sinnott gave a presentation and demonstration of one of these standards, focusing on use of VOMS attributes being pushed to services for authorisation decisions to be made on access for using Grid services protected by PERMIS. This was made in the context of the VOTES project. The pull model specification was to be explored within the nanoCMOS project; hence, the authorisation specifications mapped directly onto the requirements and scenarios in NeSC projects. On Tuesday, Prof Sinnott attended all three OGC-OGF workshops. This was a fact-finding effort in understanding what was happening in the geospatial and geospatial standards space. Furthermore, through the NeSC involvement in the SeeGEO project and the recently ESRC-funded DAMES NCeSS node in which NeSC was directly involved, knowing more about how to access and use geospatial data in social science and epidemiological studies was very much aligned with future NeSC work. Based on this meeting, initial ideas on collaboration were being pursued with some of the attendees. On Wednesday, Prof Sinnott attended the Encyclopaedia of Life presentation which covered an interesting area. Prof Sinnott also attended the workshop on financial services, an area he was not directly involved in but was interested in. He was hoping that the security technologies and scenarios on which NeSC Glasgow had been working might find a niche in this area, but the talks were concerned more with HPC and performance. Prof Sinnott attended the keynote on cloud computing. It was not clear to him how the ideas of ensembles, as presented, differed from the vision of virtual organisations in their currently supported state. That said, Prof Sinnott had a PhD student looking into configuration management and Grid technologies and considered that there might be some mileage in pursuing some investigation into this latest OGF buzz. Prof Sinnott attended the pharma workshop since it mapped directly onto numerous projects at NeSC. The panel session at the end was especially interesting. Prof Sinnott still felt that the Grid community had not yet grasped how the standards and technologies, and their deployment applied in live clinical settings. On Thursday, Prof Sinnott attended the Grid Usage and Productivity in HPC session, which was essentially a talk about Tabor Research and Grid marketing. He also attended the data management workshop and listened to talks about commercially oriented data management, including, for example, the day-to-day reality of managing the eBay infrastructures. It was interesting, but he was not sure how relevant these large-scale, industrial infrastructures were to the day-to-day activities of academic Grid-related research. Page 161 3 GridNet2 Reports 3.35 Mr David Spence – STFC All-in-all, Prof Sinnott enjoyed this OGF. As with the Seattle meeting, this was not the best attended OGF but this was discussed at length at the Town Meeting, both in terms of the financial situation of OGF and the need to avoid clashes with conferences. For example, the next OGF in Barcelona clashes with OGC meetings in Germany and with the HealthGrid conference in Chicago. However, it is co-located with the BEinGrid conference. Selecting dates that avoid clashes is a non-trivial matter given the scope of OGF and the groups and research domains it covers. 3.35 Mr David Spence – STFC The third session was taken up by a talk describing the result of the recent discussions in the Shibboleth developer community over the features and implementation methods for Shibboleth 2.0. This talk was given by Nate Klingenstein (a Shibboleth developer) and gave the Grid community a chance to see how future developments in Shibboleth would affect their work and have their questions answered. The fourth session looked at some current issues that have an effect on all Shibboleth-Grid integration projects. • Mr David Spence is involved in Grid Deployment at the STFC’s Rutherford Appleton Laboratory. In particular, he described in detail how Shibboleth can be used with VOMRS/VOMS and the opportunities for even deeper integration. For a summary of Mr Spence’s GridNet2 activities, see section 2.35, Mr David Spence – STFC. 3.35.1 GGF18 • Christoph Witzig presented discussions that he had had with the IGTF and PMAs about accreditation of Shibboleth-based CAs. • Tom Scavo presented some of the new OASIS standards which had been submitted as part of the GridShib project, especially in the area of X.509 and SAML bindings. Shibboleth for Grids The Shibboleth for Grids workshop comprised four sessions. The first two of the sessions comprised of reports on the progress of Shibboleth and Grid integration projects which were presented at GGF16: • Erik Vullings presented the continuing Shibboleth work at MAMS in Australia and in particular their IAM suite. IAM was a generalised collaboration platform allowing VO users to gain access via Shibboleth to their VO’s resources, including Grid resources in a non-VO specific way. • Christoph Witzig presented the work of the SWITCH project that was part of the EGEE-II project to enable Shibboleth-based authentication to EGEE resources. • Von Welch presented an update on GridShib. Their recent work included providing means to place SAML assertions into X.509 certificates (including OASIS standards) and integration with myVocs. • Mr Spence presented the progress with the ShibGrid project. Mr Spence covered the project’s user requirements feedback and the architecture developed. • Richard Sinnott presented various projects from NeSC Glasgow, including the DyVOSE project which was providing a dynamic privilege management infrastructure. • Mike Jones presented progress with the SHEBANGS and ShibVomGSite projects, along with questions about levels of assurance and standards for mappings between Shibboleth attributes and identities and X.509 DN identities and VOMS attributes. • David Chadwick presented GridShibPERMIS as a PDP for GT4 and the web, and recent developments in ease-of-use through a policy editor and a simplified version of PERMIS. Page 162 Alan Sill looked at issues with Shibboleth integration for registration systems, which was an important issue, especially in systems with automatically-generated X.509 identities for users. These sessions proved informative about the direction that different projects are working towards and the future of Shibboleth itself and its effect on Grid work in particular. The session showed that while many people have arrived at the same core solution for converting Shibboleth authentication assertions to GSI credentials (that is, through the use of an online CA of some description), there was not much consensus in areas such as registration, levelof-assurance and identity and attribute mapping. Due to time constraints, the discussion about an interoperability test-bed and points for interoperability were dropped; these would have been a great starting point for resolving these issues. Towards Worldwide Grid User Support This session looked at the area of providing consistent user support in Grids that span the whole globe, where users might be in a different VO, continent and Grid to the resources they were employing. This was driven by the co-operation between EGEE and the Open Science Grid (OSG), which were interoperating to provide resources for the Worldwide LHC Computing Grid (WLCG). Both Torsten Antoni from the GGUS support helpdesk in EGEE and Rob Quick from the OSG Grid Operations Centre presented their individual solutions to large-scale user support and how the two systems were currently interoperating. In both cases, they had developed similar federated approaches consisting of VO-, resource-, geographically- or function-local support units with a central system for routing non-local queries. This was a useful session looking at a non-technical issue which was having an increasing effect on Grids with growing numbers of users and resources. Although these two systems were currently interoperating, there seemed to be an outstanding need to develop standard practices (and protocols) to support increasing user levels. 3 GridNet2 Reports 3.36 Mr Anthony Stell – NeSC Security Talks Blair Dillaway from Microsoft described the company’s unified approach to trust, delegation and authorisation. This security policy assertion language (SecPAL) combines best features from other similar authorisation schemes to provide this XML-based declarative, logic-based, security language. Jeff Tan from Monash University talked about work done to circumvent institution’s firewalls to enable Grid access. For this he used a combination of SOCKS and SSH. 3.36 Mr Anthony Stell – NeSC Mr Anthony Stell is a Grid engineer at NeSC Glasgow. Mr Anthony Stell was funded by Prof Richard Sinnott’s GridNet2 award. For a summary of Mr Stell’s GridNet2 activities, see section 2.36, Mr Anthony Stell – NeSC. 3.36.1 GGF16 OGSA-Authz-WG meeting OGSA-Authz-WG In this working group session a new charter for OGSA Authz was agreed, extra features were suggested for the Credential Validation Services (CVS) requirements document, and some comments were made on the PERMIS and VOMS profiles. Mr Stell’s work in Grid research was primarily focused on security. Authorisation standards and specifications had a large impact on the technology that he used in the NeSC projects with which he was involved. IGTF Issues Grid Education and Training Workshop This session started with the introduction of the new IGTF logo and was followed by reports from the three PMAs. The draft Member Integrated Credential Services (MICS) profile, which leverages a site’s high quality accounts database, was discussed, along with changes to the Classic CA profile. This part of the session led to a discussion of the relative roles of the IGTF and PMAs. In the second part of this session, Peter Alterman gave a presentation on the Federal (US Government) PKI Architecture. This talk especially focused on PKI trust bridges which allow trust to be bridged between different trust providers. This talk proved useful in thinking about the future of Grid trust. FI-RG In this session Melinda Shore from Cisco, who is also chair of the MIDCOM-WG at IETF, presented the various solutions to firewall issues to kick off the next stage of work in the FI-RG. The next stage was the the evaluation of the IETF solutions to these problems and the creation of a solutions document. In addition, other solutions and the subsequent discussion were presented and led by Ralph Niederberger, Thijs Metsch and Jeff Tan. This included ideas about how to enable firewalls for Grid usage. Topics in Identity Management The Topics in Identity Management session consisted of three talks: the first was by Von Welch who gave an overview of GridShib, and the second and third were given by Stephen Langella (Ohio State University) and described Dorian and the Grid Trust Service (GTS) respectively. These were components which provided authentication, authorisation, trust management and secure communications services to medical collaborations. The Dorian system provided the Grid user account management function and GTS provided secure inter-institutional trust. These talks were useful case-studies in different methods for providing easy-to-use secure access to Grid resources for non-Grid users. Part of Mr Stell’s remit at NeSC in Glasgow was to assist in the teaching of the Advanced MSc Grid Computing module. As such, his attendance at this session aided the discovery of other methods of teaching Grid and discussion of the issues raised in this field. Shib and Grid Investigators Meeting The security work that formed the mainstay of Mr Stell’s research work largely involved Shibboleth and other technologies that interface with it. Attendance at this session gave him an overview of other projects that were investigating Shibboleth at the time and how these related to the projects in the same area that were being conducted at Glasgow. GT4 Status and Experiences Mr Stell was using GT4 as part of the development of the VOTES project and had previously been involved in the alpha-testing stage of GT4. Mr Stell attended this session to gain an overview of how v4 of the toolkit had since progressed and to discuss whether new features have appeared that might help with current development efforts. OGSA-DAI Technology Update As with GT4, Mr Stell was using OGSA-DAI in his current project and an update in the status of the technology was useful for knowledge of its implementation in the VOTES project. Security and Privacy Needs of Health Grids This meeting had a direct effect on work in Mr Stell’s current project as this was exactly the application domain that is covered by VOTES, with a focus on the security aspects that are prevalent in the life sciences. Grid Authorization Interoperability Workshop The issues that were appearing in the establishment of a Grid authorisation infrastructure also had a direct effect on previous security work and the issues that were prevalent in the VOTES project. Page 163 3 GridNet2 Reports 3.37 Dr Ian Taylor – Cardiff 3.37 Dr Ian Taylor – Cardiff Dr Ian Taylor is a senior lecturer at the School of Computer Science, Cardiff University. For a summary of Dr Taylor’s GridNet2 activities, see section 2.37, Dr Ian Taylor – Cardiff. 3.37.1 GGF18 Dr Taylor participated in the following sessions at GGF18: • Co-organised, chaired and gave introductory comments at the Appliance Aggregation Research Group (APPAGG-RG) where the theme was the integration of static and mobile Grids. • Attended a number of other sessions and keynotes, including the SAGA working group sessions, the EU funding round table, the OGSA-WG and the LSG-RG, amongst others. APPAGG-RG The APPAGG-RG session was called Aggregating Mobile Devices with Grids. It attempted to focus on the co-existence of static Grids and mobile Grids by extracting use cases in that area. Dr Taylor coorganised the session, with Vladimir Getov. The agenda was as follows: • Co-organiser and Speaker of the Sharing Workflows session (WFM-RG) • Participant: OGSA Workflow • Participant: e-Arts and e-Humanities e-Science: Technologies and Methodologies in Arts and Humanities Research (two sessions) • Participant: SOKU Workshop Sharing Workflows Session Dr Taylor chaired and co-organised the Sharing Workflows session with the group’s co-chair, Ewa Deelman. The session focused on the need for sharing workflows for scientific applications, rather than focusing on standardisation of workflow representations, which other groups, such as the OGSA Workflow group, are looking at. The session had a number of speakers with a range of topics within the workflow-sharing paradigm. The agenda was as follows: • Dr Ian Taylor: Introduction and Motivation • Adrian Toth, University of Miskolc: Levels of the Grid Workflow Interoperability • Omer Rana, Cardiff University: Workflow Optimization and Sharing Using Performance Information • Rizos Sakellariou, University of Manchester: Scheduling Data Intensive Workflows onto Storage-Constrained Distributed Resources • David De Roure, University of Southampton and OMII-UK: myExperiment - Social Software for Workflow Sharing • Andrew Harrison, Cardiff University: The WHIP Plug-in for Workflow and Artefact Sharing • Richard Sinnott, National e-Science Centre, Glasgow: Shibboleth Protection and Management of Workflows • Dimosthenis Kyriazis, Telecommunications Laboratory, National Technical University of Athens: A Workflow Mapping Mechanism for establishing Quality of Service Guarantees • Discussion on the Focus Research Document on Application Scenarios for OGF-21 1. Dr Ian Taylor (Cardiff University) , Session Introduction 2. Vladimir Getov (University of Westminster, UK), Time-Critical Use Cases in Mobile Grids 3. Brian Adamson (Naval Research Lab (NRL), Code 5522: Composite Computing and Mobile Ad-hoc Networks 4. Ian Wang (Cardiff University), P2PS: P2P for ad-hoc Networks and Simulation 5. Adina Riposan (Contact Net, Romania), Mobile Devices and eHealth Paramedical and Imaging Scenarios 6. Omer Rana (Cardiff University, UK), Healthcare@Home: Integrating Healthcare Sensors with Grid-based Data Analysis The attendance was around 20 people so some interest was shown within the community, but Dr Taylor felt that it was not enough to drive this theme into the following OGFs. Dr Taylor could not attend OGF21 because he was teaching at Cardiff—the OGF was held on the open week of the semester—and so the momentum for the APPAGG-RG dwindled. A few months before the OGF in Manchester, Ewa Deelman suggested that Dr Taylor became the co-chair of the WFM-RG instead, which was timely because Dr Taylor had just edited a special Journal of Grid Computing Journal on Workflow, had been lead editor on Springer’s Workflows for e-Science and had a number of talks with other workflow groups. The WFM-RG therefore became Dr Taylor’s focus for OGF20 and onwards. 3.37.2 OGF20 Dr Taylor participated in the following sessions at OGF20: Page 164 The session was a success with a number of discussions developing from the various presentations. The speakers and participants were enthusiastic about being involved in the creation of an OGF research document on the subject of gathering use cases for sharing workflows. This would be initiated with progress made in time for OGF21, where the RG would discuss the use cases and take the subject further. OGSA Workflow As co-chair of the WFM-RG, Dr Taylor attended this session to observe current work in the workflow arena for OGSA. The general theme of the group was to gather current use cases and discuss possible standardisation of workflow for OGSA. There were a number of discussions about BPEL4WS and its suitability for OGSA. 3 GridNet2 Reports 3.37 Dr Ian Taylor – Cardiff There were also discussions about the Workflow Management Coalition (WFMC), a global organisation of adopters, developers, consultants, analysts, as well as university and research groups engaged in workflow and Business Process Management Software. The discussions ended up in favour of BPEL to some extent, but both the discussions and use case gathering were at a very early stage. Dr Taylor planned to monitor this group over the coming OGFs to see where they would go with this and whether there would be adoption of a specific workflow language for Grid computing. Most current workflow systems had not adopted such an approach so it would be interesting for Dr Taylor to see how the field evolves over the coming years. such interoperability and if so, how to achieve this. The agenda was as follows: • Dr Ian Taylor – Session Overview • Matthew Shields – Sharing Workflows Recap of OGF20 • Volunteers for the research document – Use Cases for Workflow Sharing • Ewa Deelman – NSF Workflow Interoperability workshop and what transpired in the execution breakout • Andrew Harrison – Kepler/Triana Integration Interoperability Use Case 1 • Maurizio Melato – Web-Based Grid Workflow System: A-WARE Project • David De Roure – myExperiment • All – Future plans (Use case document) e-Arts and e-Humanities This workshop was a BoF split into two sessions, with presentations from the work of TextGrid in Germany, the Arts and Humanities e-Science Initiative in the UK (David De Roure) and presentations from related projects in the US – Steve Beck was amongst the speakers. The second session involved more discussions about possible routes to standardisation in e-Arts and e-Humanities. One of the presentations (Steve Beck’s) included some slides on the work being done in Cardiff in this area, specifically on the DART mini-PIPSS PPARC project, which was looking at using P2P and workflow technologies for distributed music information retrieval. SOKU Session An underlying theme in a number of OGF sessions was the increasing focus on Web 2.0 and social network-based technologies to support collaboration between application users. Web 2.0 technologies (such as the AJAX, JSON, etc) and mash-ups were a hot topic in the SOKU session also. The session included presentations from a number of speakers including David De Roure and Pinar Alper from the UK. Dr Taylor attended the session out of interest and to understand the different perspectives on SOKU. It was interesting to see the vision from the different speakers. 3.37.3 OGF21 Dr Taylor participated in the following sessions at OGF21: • Co-organised, chaired and gave introductory comments at the Workflow Management Research Group (WFM-RG) session. • Co-organiser and participant of the GridNet2 e-Science Workshop (three sessions). • Attended a number of other sessions and keynotes, including Think Little: The Proliferation of Small Clusters Means Big Changes, Web 2.0 Meets Grids panel, e-Social Science: ourSpaces amongst others. • Held a number of follow-up meetings for the WFM-RG Session. Dr Taylor chaired and co-organised the WFM-RG session with the group’s co-chair, Ewa Deelman. The title of the session was Workflow Sharing and Interoperability and focused on extracting some user requirements from existing systems for being able to interoperate between current workflow environments. The group discussed and tried to answer questions such as whether there was a call for The session was timely, appearing just after a Workshop on Scientific and Scholarly Workflow Cyberinfrastructure: Improving Interoperability, Sustainability and Platform Convergence in Scientific and Scholarly Workflow 11. The session carried on with the themes from this workshop (which both Deelman and Dr Taylor attended and gave presentations at) in order to get feedback from the OGF community in the usefulness of such interoperability. The results were promising. The session was a success with around 40 participants and several people made comments at the closing discussion about their experiences and noted that they would be willing to provide use cases. This work was complementary to other effort in this field. A networking proposal W4eSNet: Workflows for e-Science Network was to be submitted to the e-Science networking call with many participants from this session and the field in general. The WFM-RG were hoping to build up momentum in OGF22 on this theme by gathering use cases and presenting these at the next OGF. When the use cases were complete, the group might choose to propose to spin off a working group in order to create standardised interfaces for interoperability between workflow systems. 3.37.4 OGF22 Dr Taylor participated in the following sessions at OGF22: • Co-organised, chaired and gave introductory comments at the Workflow Management Research Group (WFM-RG) session. • Attended a number of other sessions and keynotes. • Held a number of follow-up meetings for the WFM-RG Session including interactions with other groups, for example, OGSA-Workflow. This report focuses on the ongoing activities within the WFM-RG by providing an overview of the session 11 Supported by NSF and the Mellon Foundation. October 4-5, 2007 Page 165 3 GridNet2 Reports 3.38 Mr Elias Theocharopoulos - NeSC and the current direction of the RG and how this will lead to standards activities over the next year. Dr Taylor chaired and organised the WFM-RG session. 3.38.1 OGF22 Mr Theocharopoulos attended the following meetings at OGF22: • Opening Session • SAGA Java Language Binding • OGSA-Authz-WG • WFM-RG • OGSA-Data Architecture Future Directions • OGC-OGF Collaboration Workshop • Infrastructure Management – eBay • SAGA, the Simple API for Grid Applications • SAGA + DAIS: Next Steps • The Encyclopedia of Life: A Web page for Every Species The WFM-RG session was very well attended with 34 registered participants. The discussion and presentations focused on three levels for interoperability and tried to get people to comment on these. The approaches discussed were: • Porting Applications with Globus Gridway • Data Management Workshop • Workflow embedding – allowing workflows to run within their own environment, but invoked from another. • Development of a meta language – allowing different proprietary languages to be mapped to a single standard one. • Semantic annotation/description/classification – this is particularly important for sharing. Mr Theocharopoulos’ principal role at OGF22 was to present his work on an implementation of WS-DAI and, more precisely, of DAIX using OGSA-DAI. Therefore, he participated in the DAIS-WG session by giving a presentation about the implementation status and issues raised. Providing the status of this effort was important for the next steps of the DAIS-WG and was highly appreciated by the authors of this specification, highlighting possible errata. As a result of Mr Theocharopoulos’ contributions, there were discussions planned in order to get certain parts of the WSDAI/DAIX documents clarified. In preparation for the session, Andrew Harrison and Dr Taylor had written a questionnaire, for which Mr Harrison created an on-line submission form (http://bender.astro.cf.ac.uk/wfmrg), to gather opinions, experiences and requirements from users, designers and developers of workflow systems regarding interoperability and sharing. The questionnaire was sent to a number of external mailing lists, and also to internal ones, such as the WFM-RG and OGSA-WG. Dr Taylor and Mr Harrison received 14 responses before the session. The session looked at sharing and workflow interoperability and led on to a discussion of the survey results, in an attempt to get volunteers for contributing towards a research document. Dr Taylor gave the introduction and session overview, including background to the research and past efforts in interoperability. The discussion proved very successful and seven people volunteered to help with the new research document. Work on the document began and, since the session, another two people volunteered to help. Dr Taylor hoped that this research document would bring together previous efforts on workflow interoperability (for example, from the WFMC) and create a taxonomy for future discussions on interoperability. Depending on the success of the document and the feedback from the community in future OGF sessions, this work could lead to a working group being established in order to help provide standardisation for workflow interoperability in general. The aims of the WFM-RG session were fully met and the group was very happy with the result and enthusiasm from the community in taking this work forward. 3.38 Mr Elias Theocharopoulos NeSC Mr Elias Theocharopoulos is the DAIT Software Engineer at the National e-Science Centre in Edinburgh. Mr Elias Theocharopoulos was funded by Dr Steven Newhouse’s GridNet2 award. For a summary of Mr Theocharopoulos’ GridNet2 activities, see section 2.38, Mr Elias Theocharopoulos – NeSC. Page 166 DAIS-WG The main points to be discussed were as follows: • Consider JDBC constraints in order to get full or intermediate tuples back, as well as getting number of rows back. • Consider clarifying errors in DAIX in a more precise way, that is, give more information to the client. • Consider re-phrasing parts that sounded ambiguous to the implementers. The benefits of Mr Theocharopoulos’ attendance at the event included meeting with the authors of the WS-DAI specification and others from the DAIS-WG—Isao Kojima and Said Mirza from AIST, Japan, and Oscar Corcho and Miguel Esteban from UPM, Spain—who are working on new realisations for RDF to exchange ideas and experiences of implementing the WS-DAI family of standards. Theocharopoulos planned to continue to be involved in the ongoing discussion to resolve issues and inconsistencies raised by his implementation. Mr Theocharopoulos discussed the following issues with Miguel Esteban: 1. Investigating whether his DAIX implementation could support some advanced XQueries 3 GridNet2 Reports 3.39 Dr David Wallom – OeRC 2. Esteban’s concerns about the response of the factory operations, and more precisely, the fact that they seemed to be inconsistent with the core specification • Requirement: Embed existing workflow that run in a different environment • Challenge: Define workflows in an abstract way. Mr Theocharopoulos spoke to Mike Jones from Manchester about his work on the SARoNGS project, which is, effectively, a gateway to use Shibboleth authentication in order to grant authentication from NGS. • Benefit: Re-use • How to achieve it: OGC-OGF Collision Workshop Mr Theocharopoulos attended most of the OGC-OGF Collision workshop. The aim of the workshop is to bring the OGC and OGF communities together. This involves: • Define workflow at abstract level • Allow workflow engines to be sub-forked as a sub-process. Some interesting questions raised were: • Whether to embed locally or in a distributed sense. • Enriching OGC's WPS with distributed Grid mechanisms • • Providing workflow functionality in order to get the data from WPS and the relevant services (WMS, WFS, WCS) Native engine enactment of shared workflows (embedding) or shared abstract (meta) description. • There had not been any major progress towards this but it seemed that BPEL was the most accessible approach on first instance. How to approach abstract meta-language, semantics: either global solution, or small things on a project-to-project mapping basis. • Is it worth attempting a classification of workflow requirements? • How to deal with the increasing complexity due to the changes of the technology to update the versions of workflow engines. There was also some discussion about adopting SAGA as an API to build geo-spatial applications. There was a need for some level of authentication especially when demographic data was involved, because of its sensitive nature. The consensus was that an external authentication mechanism/service should take care of the authentication, for example, Shibboleth. There was a talk from Muenster on the WPS-G(rid) implementation, which was based on North 52 project. In this case, they used WAS (Web Authentication Service) for user authentication, by checking the SAML credential, and WSS (Web Security Service) as a layer to prevent access to the Web Service. They also emphasised that they cared mostly about getting their work done efficiently rather than focusing on standards. Mr Theocharopoulos’ general impression was that there were a number of interesting efforts in the area of GeoSpatial services for both workflow and security directions. It seemed, though, that bringing them under the umbrella of a common standard might not be a straightforward process. WFM-RG Mr Theocharopulos attended the WFM-RG out of personal interest. The topic discussed was how different workflows that apply in Grid environments could possibly be unified, or how one workflow could use another in an efficient and interoperable way. The different types of interoperability were mentioned at the beginning. These were as follows: • Direct • Message passing • Bridging • Use of shared data stores Then the challenge of interoperability between workflows was presented, which can be summarised by the following points: The OGSA-DAI/Taverna endeavour in OMII-UK may be worth contributing as an example of interoperability to this group. Theocharopoulos has passed on the contact details of the people who implemented this work to the group chairs. From a personal point of view, Mr Theocharopoulos improved his presentation skills and was exposed to work in other groups, in particular the workflow management and SAGA groups which would be of use to his future work. 3.39 Dr David Wallom – OeRC Dr David Wallom is Technical Manager of the Oxford e-Research Centre. For a summary of Dr Wallom’s GridNet2 activities, see section 2.39, Dr David Wallom – Oxford. 3.39.1 GGF18 Dr Wallom attended the following sessions at GGF18: • GGF Opening, Mark Linesch • Keynote: Vision for 21st Century Discovery, Dr Dan Atkins, NSF • GIN Update • Topics in Grid Management Workshop (Chaired) • PGS-RG Workshop Followup and Planning (Chaired) • UR-WG Document discussions • OGSA HPC Profile discussion Page 167 3 GridNet2 Reports 3.39 Dr David Wallom – OeRC • Daonity: Trusted Computing Enhanced GSI Workshop • Chairs Update • Storage Grids in Healthcare Workshop GIN Update The GIN group gave an update on their current status. The Authentication and Data groups were well along the path with targets being met. The Information group gave a presentation a graphical tool that presented GLUE information. • • PGS-RG The PGS group made several recommendations on for what they would do over the next 2-3 meetings. This included a wrapup of the workshop that the group had held previously. UR-WG Document Discussion This session gave an update on the changes that were received through the period of public comment and that had been acted upon since GGF17. OGSA-HPC Basic Profile Software Licensing for Grids BoF Dr Wallom was co-organiser for this session. • Grids Mean Business: Software Licensing - A Way Forward? • UR -WG • The Astronomical Virtual Observatory Building Operational Services on Pervasive Grids: Standards in Use • Data Management Area • Experience with Application Domains, Massimo Lamanna (CERN) Topics in Grid Management This workshop bought together all the managers of different production Grid infrastructures so that they could describe the best practice examples that could be used in the new and upcoming Grid infrastrucutures. This workshop attracted a large number of attendees and was considered a great success. The Grid as a Complex System: Faster, Bigger, Better Science on Global Grid Infrastructure, Peter Coveney 2nd International Workshop on Campus and Community Grids This workshop was massively attended, with the conveners having to turn people away from the room as people were already sitting in the aisles. The aim of the workshop was to illustrate that for each of the topics that were discussed within the breakout sessions, a Grid of any size, be it organisational or international actually faces the same problems. From this, the group aimed to give a lead to the further topics that should be touched on by GIN as it moved from a technical exercise to a usable long term infrastructure. The output from this workshop was a published OGF informational document. At this meeting, there was an update on the status of the HPC-BP work, which included the number of groups that were going to implement HPC-BP when it was finalised. These groups included Platform, MS and possibly some of the open source solutions. The workshop closed with a panel discussion and round-up of the breakout sessions. Daonity: Trusted Computing Enhanced GSI (Workshop) • OxGrid and the UK NGS • GLOW and US Open Science Grid, Wisconsin • White Rose Grid • Crimson Grid, Harvard • Cardiff Campus Grid The Daonity product from HP (China) is a trusted computing implementation. This session gave an overview on what had been done so far. Storage Grids in Healthcare (Workshop) This session was a requirements gathering effort. Dr Wallom gave a description of the eDiamond and future GIMI infrastructures. 3.39.2 OGF20 Dr Wallom attended the following sessions at OGF20: • 2nd International Workshop on Campus and Community Grids This continued the interoperability theme and was an all-day session. • Keynotes: • The Social Grid, Tony Hey Page 168 Campus Grid Case Studies: Connecting Local Services to a Wider Context Community Grid Case Studies: Connecting Local Services to a Wider Context • eMinerals • NEESGrid (or UK NEESGrid) • Data Grids: • ORION • BIRN • European Data Grids 3 GridNet2 Reports 3.39 Dr David Wallom – OeRC National General Infrastructures: Federating Smaller Grids This session looked at specific mechanisms for engaging infrastructures with different software stacks. • DGrid • EGEE • Naregi • NGS-UK • OSG • TeraGrid Breakout Sessions • User management, passing identities and accounting • Support models, both systems and users • Software Licensing within cross organisational systems The results of this session were used to lead into the discussion in the Grids Mean Business session later in the week. • Data in Grids The key topics included authenticity and integrity, access controls, and technology evolution management. The Social Grid Tony Hey spoke about the importance of knowledge management and used many examples of institutional repositories etc. The Grid as a Complex System: Faster, Bigger, Better Science on Global Grid Infrastructure Peter Coveney spoke about Reality Grid and his other wide area projects, introducing the necessity for higher level functionality, such as computational steering and so on, and how it could be ensured that standards are developed in the higher functionality areas as well as the lowest levels. Software Licensing for Grids BoF This was a summary session for the round table that was held at SC06 at the UK e-Science booth. This round table session looked at the issue of using licensed software within a Grid environment, an issue that had arisen in previous meetings of the PGS-RG at GGF/OGF events. The session covered the user issues and several companies were invited to contribute vendor issues. Conclusion and Way Forward It was decided that the best way for would be for those involved to try to document the use cases that showed how their software had been used within a Grid environment. This meant little extra work for the commercial organisations, but it was hoped it would be of great benefit to users. The model the PGS-RG would hope for is where, for example, the company had a section on approved license models within Grid environments in its support FAQ. For example, there could be short write-ups based on answers already given by support or sales staff to questions asked by current Grid users. Such FAQs could then be referenced from within the Production Grid services website, as well as from National Grid projects sites. Grids Mean Business: Software Licensing - A Way Forward? Chaired by Ian Osborne, from the Grid Computing Now KTNs, Osborne first gave a presentation on why licensing is an issue from the industrial point of view. Dr Wallom gave the second presentation, giving a summary of the PGS Licensing session from the previous day. This was intended to summarise the number of different issues that had been identified. The output from this session was an agreement that a number of those present would agree to construct a set of model use cases which would be able to describe licensing in a Grid environment. Users (both commercial and academic) could then use the resource to negotiate with suppliers. It would show how other companies had worked out a suitable licensing model. UR-WG Following on from the release of the version 1 standard into public comment for the second time, this session was used to sound out how the group would consider extending it towards version 2. It had been shown that although it was good for computation, there was a lot of extra functionality needed for storage, networking and, overall, a service. The Astronomical Virtual Observatory - Building Operational Services on Pervasive Grids: Standards in Use Dr Wallom attended the final two sessions of this workshop, but this meant that he did not get too much specifically from the talk. Dr Wallom had been asked to be on the panel since he was involved in production Grids and production issues. Data Management Area Dr Wallom attended the first three sessions of the Data Management Area discussion, but he found that, even though this was supposed to be an EGEE User Forum session, these sessions were dominated by infrastructure people. This clearly highlighted the difference between any users other than particle physics users and the EGEE infrastructure developers. Experience with Application Domains This session was much more application-focused and useful. It did show the number of user areas that were working on EGEE. 3.39.3 OGF22 Dr Wallom attended the following sessions at OGF22: • Plenary Opening Session Craig Lee, OGF-Europe, David Wallom Page 169 3 GridNet2 Reports 3.39 Dr David Wallom – OeRC • OGF-Marketing • OMII-Europe • Grid Information Retrieval-WG • Plenary: What OGF Can Do for Enterprises – A View from the CIO Office, Charlie Catlett • HPC-BP • GIN-CG • Town Hall (two sessions) • Creating a Standard Software API for Data Grid Management Systems BoF • The Encyclopedia of Life: A Web Page for Every Species • Financial Services Workshop (second session) • Plenary: Cloud Computing, Grids and the upcoming Cambrian Explosion in IT, Irvine Wladawsky-Berger • Cloud Systems BoF • Data Management Workshop (all-day sessions) OGF-Marketing Walter Stewart chaired the meeting and noted that until the outputs from the Tiger Team became clear, a marketing strategy change would be on hold. Paul String from the Tiger Team presented some of their findings; the main one is that OGF should re-focus on its more traditional e-Science/community background. This was to include the development of a marketing strategy that focused on ensuring that the key successes of the organisation are heard about outside. There was also the feeling that the marketing activity within the OGF had dropped in level of importance and, as such, should be part of the GFSG meetings. OMII-Europe Steve Brewer gave an overview presentation of OMII-Europe and its current status. OMII-Europe was due to finish soon. It was not clear from this what the concrete outputs from OMII-Europe were. Alistair Dunlop gave an overly rosy summary at the end of the talk. The Italian group (represented by Sergio Andreozzi) gave an overview of the progress to standardisation of GLUE version 2. Dr Wallom noted that there was a distinct lack of people from the NGS working on this standard. The plan was for the current standard to go into public comment in the weeks after OGF22. Andreozzi then gave a description of the outputs of GLUEMan, which is a system that can collate and format the outputs from a GLUE information source into different output ‘renderings’. Morris Reidel gave an overview of a project that he had been running which uses resources within EGEE and DEISA. This was a biochemistry application, with trivially parallel pre-processing to localise protein docking sites of interest, and then more focused work is done on best candidate sites on one of the DEISA Page 170 resources. This was a very interesting example of multi-Grid use. Grid Information Retrieval This session was run by Greg Newby and was predominantly about future directions. Plenary: Charlie Catlett Charlie Catlett gave a talk showing how the standards, best practices and so on that are recommended by the OGF must be focused towards ease of implementation and usage. This could include how cloud computing is used in the future, and so on. HPCBP Specification Adoption This session had participation from MS, UVA and Platform. The conclusions were that this very important group and output from the OGF is in good health, with a number of implementations and demos from other areas. Dr Wallom asked whether they have heard if anyone is implementing HPCBP as a base service on a cloud network. There was going to be a SAGA/BES implementation available though Vrije University. GIN-CG Morris Reidel was introduced as the group’s new cochair. He gave a presentation on the status of all the participating Grids and the standards that they were using currently and that they planned to use in the future. This included Dr Wallom’s input for the NGS. Dr Wallom also pointed out that now that there was this physical infrastructure linkage, it was really necessary to show a user requirement. This would also highlight the necessary management structures that would need to be in place to do cross-Grid accounting and so on. Town Hall As a representative of OGF-Europe, Dr Wallom was asked to be available to answer questions on the project. In the end, the major questions were on what the status of OGF would be over the longer term due to some sponsors cancelling their participation and so on. This was answered with calls about the need for increasing focus and to start to work towards capturing application vendors rather than hardware vendors, because if OGF work was going to be more user-focused, then the majority of users were not interested in that lower level but would be interested in how, for example, they would be able to connect to a Grid directly from within an application such as GAUSSIAN, for example. Other questions concerned the NOMCOM, which was still looking for area directors. OGF/EGEE Users Forum Procedures Dr Wallom attended a meeting with EGEE and OGF management on procedures for the spring OGF/EGEE Users Forum and how the Requests for Proposals to host needs to be decided and so on. This included suggestions for locations and minimum requirements. The EGEE and OGF RFP were both passed on to Silvana. 3 GridNet2 Reports 3.39 Dr David Wallom – OeRC Creating a Standard Software API for Data Grid Management Systems BoF This BoF was chaired by a member of Reagan’s team, and was poorly presented, with many participants assuming that the description meant that the BoF would just be about extensions to the file handling capability of SAGA. Afterwards, Dr Wallom realised that it was subtler than that, and was about not only the ability to separate the logical and physical file locations but also, of course, have metadata and separated data sets. The Encyclopaedia of Life: A Web Page for Every Species Jennifer Schopf gave a roundup of her current position within the Woods Hole Oceanographic institute where she is working on the Encyclopaedia of Life (http://www.eol.org). The aim is to provide a single page (with user controlled detail levels) for each of the ~1.8M different species of flora and fauna. The site went live one week before OGF and died within 11 hours due to extensive media coverage! It would appear that this was very much a proprietary system with little in the way of real-time data interoperability. It was also intending to cache all of the information in a single location, with little or no requirement for data linkage between different databases. It was interesting, although not too relevant to OGF. do with the system. If, for example, they were trying to have a number of interacting services on resources that they have rented and they can end up on multiple different servers each time they are instantiated, their overall performance may differ greatly. There was, at the time, no group that was dealing with the necessary standardisation of interfaces to cloud systems, the closest being the Computing Community Consortium who are organising events to try to get the cloud community together. This was therefore a green field site for the OGF, who had the expertise to show how interfaces defined within OGF can be used to access clouds. There are several technical issues that must be addressed: • Performance overhead in terms of matched CPU performance? • Cost gains, including size efficiency which may include green considerations. • Security had currently not been fully explored and so there was no clear illustration of what levels of user-provider privilege separation exist. • What is the model of computing can you provide? What type of internal configuration is available? With suitable network connectivity between nodes, for example, can you do capability computing in clouds? • With major scientific problems, there is normally a need for data-compute affinity due to the sheer volume of data you need to process. • Is there an underlying scheduler for the use case where demand exceeds supply? How would this affect externally available services? • Is there going to be a situation where users would want to link clouds together? For example, would a user want to equate usage between different suppliers of cloud services? • Are individual institutions or groups going to want to construct their own clouds, as they have campus Grids? Financial Services Workshop Dr Wallom attended the second part of the Financial Services workshop, which consisted mostly of presentations from providers of solutions including Platform, HP, and IBM. It did not seem very relevant to the financial services community unless they didn’t know already what was available to them in the way of systems. Dr Wallom again met representatives from Cycle-Computing, as he had at the HTC workshop in Edinburgh. Cloud Computing: Grids and the upcoming Cambrian Explosion in IT Irving Wladawsky-Berger highlighted that the direction in which OGF is currently heading woyuld undoubtedly further abstract the underlying computing infrastructure from the user, and even in some cases, the direct user-contactable provider of service. Wladawsky-Berger did not mention Microsoft in his presentation. Cloud Computing BoF This BoF was extremely well attended with upwards of 60 people in the room. The meeting was divided over two sessions, with two talks in each session and a discussion at the end of each. Geoffrey Fox, who was defining the problem space from the viewpoint of the OGF, started the meeting. What are Clouds? Fox highlighted that in a lot of ways Grids and clouds are interchangeable in their overall end goals, though they intend to ensure that the users of the system are more insulated from the internal workings of the cloud. Dr Wallom noted that this could have issues depending on what the users were actually trying to It had also been shown that interoperability was necessary for these systems, as, for example, when the Amazon S3 system went down for a period recently. This could have catastrophic consequences for a business/project that had this as a core part of their business. The point was made that clouds could give MSI colleges and other education institutions throughout the US access to computing resource they normally would be able to muster. Dr Wallom noted the group should ensure that the Clemson example from HTC week gets to Geoffrey since this is a great example of a university using Grids to get computing resources they didn’t have access to before. The final question from Geoffrey was should the OGF have a cloud computing group? The next speaker was Steven Newhouse, who questioned severely whether clouds were new or just a logical extension of the methods of making computing available to users from over the last 30 Page 171 3 GridNet2 Reports 3.40 Dr Viktor Yarmolenko – Manchester years. He again highlighted the need for standards at two levels: • • Infrastructure: low-level compute and data, as well as management and monitoring Interface services: AAA, as well as functional applications and domains With standard APIs needed, for both the development of the cloud system itself, as well as the higher-level services that the providers decide to run themselves, the question from Newhouse is what can OGF do? The obvious next step is, for example, to show connection to a cloud instance with a light weight interface between the cloud service interfaces and OGF standardised interfaces. The third talk was by a representative of Internet2 who started by questioning the effect that cloud could have on the existing network infrastructure. The biggest step would have to be in the perception of the network within the whole infrastructure. It must become an integrated part of the whole to use clouds efficiently and reliably. The transparency of commercial networks would also become a problem when trying to route as efficiently as possible between different compound services. This would also therefore have to include smarter network management to take into account the changes to traffic characteristics. Dynamic circuit networking would change the paradigm of what was considered to be ‘the network’. Geoffrey Fox gave the final talk on behalf of Kate Keahey, Argonne Nat Labs, who had been doing initial testing on use of cloud systems. This testing used two different applications, the first of which was a HEP application, STAR. This is a complex application with ~2M lines of C++ and FORTRAN developed over 10 years. It has extremely stringent requirements on the environments in which it will run, and it was becoming very difficult to guarantee the availability of such environments at all participating institutions. Through the provisioning of a virtual machine built to these requirements, Keahey has been able to study performance and so on, and shown that it is slower than a directlyinstalled instance of the application, but it has a narrow delta in performance. Dr Wallom noted that this was probably due to network variability. Keahey’s second application was a glide-in type of application that takes over any resource, Grid or cloud, and so on, to become part of her system. Dr Wallom noted that this was the same as Condor glide-in and that it seemed a bit of a waste of time to re-implement it all over again. The overall feeling from the meeting that a workshop at the next OGF would be essential, with various suppliers of cloud resource invited as well as the Trusted Computing Group. The formation of a community group though should wait until after the workshop, though OGF was starting a trawl for cochairs, in which Dr Wallom was interested as the PGS-RG seems to be dying a natural death. Plenary: Grid Usage and Productivity in HPC Addison Snell, VP/GM of Tabor Research, presented the results of a survey on Grids, HPC and utilisation. Page 172 Data Management Workshop Session 1 Paul Strong talked about eBay, which was a truly massive scale problem, far beyond just keeping it working in the base case. There were interesting anecdotes about how even plug & play systems give enough of a system interrupt storm that it can backup the eBay system! Erwin Laure, the Area Director, gave a roundup of current status. There were lots of tools for file movement but not that many for data movement. Session 2 The Storage Network Industrial Association is moving from just the underlying storage to focus more on information and data management. This is very connected to the OGF mission. David Black, SNIA Tech, gave a very enthusiastic description of the XAM interface that gives a standardised layer between data (including metadata) and client applications. This is a very similar idea to SRB, but at the hardware level. Session 3 Jens Jensen gave a description of what SRM is. Session 4 This covered GridFTP and a description of the latest Globus implementation. Andrew Grimshaw described an implementation of gridftp backend with a file system front-end using RNS, Genesis II. This looks quite interesting and could be followed up with a test deployment. Overall, Dr Wallom thought this was a useful workshop, though the point was raised there were no users present, other than HEP, and another workshop should be held in Barcelona, perhaps called Data Management 2, Revenge of the User! That way, the group might be able to steer the development of some of the higher-level interfaces such that users could interact with the systems in very much the same way that they interact with their desktops currently. 3.40 Dr Viktor Yarmolenko – Manchester Dr Viktor Yarmolenko is a research associate at the School of Computer Science at the University of Manchester. For a summary of Dr Yarmolenko’s GridNet2 activities, see section 2.40, Dr Viktor Yarmolenko – Manchester. 3.40.1 SLA F2F Dr Yarmolenko attended various F2F meetings in Germany in order to discuss SLA issues. Background Dr Yarmolenko’s participation in OGF started at OGF10 in Berlin, March 2004 where he presented his work on support for dynamic workflows (pre SAGA). 3 GridNet2 Reports 3.40 Dr Viktor Yarmolenko – Manchester Ever since he began working on Service Level Agreements (SLAs) in early 2005, he observed a growing interest in SLAs in the context of the workflows. Initially, SLAs were only considered in the context of independent jobs. This interest grew even at a faster rate between the time when he chaired a meeting on SLAs for the workflows (as a part of a CoreGRID WP6 meeting 12, co-allocated with OGF20 in Manchester) and the present day. Despite a growing interest in these two areas, the current standards largely ignore workflows; in particular, the WS-Agreement standard does not touch on workflow-related aspects of SLAs. In view of this situation, Dr Yarmolenko arranged several meetings with OGF members and commercial companies that are particularly active in the work on Grid standards in the area of SLA and are interested in its practical applications to the workflows. The meetings took place in Karlsruhe University (11-12 March 2008) and Dortmund University (13-14 March 2008) in Germany largely because the majority of participants were present at these locations at the same time. The suggested agenda was to discuss and mainly bring up his concern regarding the lack of support for workflows. At the meetings, there were discussions about several aspects of state of OGF standards and their support for Grid workflows. These are encapsulated in the following points: 1. The advantages of dynamic SLAs 13 and their potential in solving the workflow support problem. 2. The ability of WS-Agreement documents to refer to other WS-Agreement documents and how important and or necessary this is for workflow support. 3. The ability of other parties (besides initiator and receptor of WS-Agreement) to renegotiate parts of the agreement (likely with limited access). 4. What are the other standards that may be affected by and involved in the support provided for workflow management on Grid? Dr Yarmolenko’s Contribution Dr Yarmolenko suggested the following scenario to consider when addressing the points which he outlined in the agenda (above). The workflow allocation and execution cycle on a Grid system is depicted in the figure below. The client submits a (computational or other) workflow to the workflow execution service (FW Exec), which schedules parts of the workflow to the local resource managers (LM) or other FW Exec (see the diagram). Each arrow in the diagram would require a separate SLA to be negotiated and created between the two relevant parties for every segment of the entire workflow submitted on a distributed resource. In addition, each arrow may be crossing an interface between different administrative domains. The efficient scheduling of a workflow or even general robustness of the system decreases dramatically with the scale of the system. In such scenario, a flexible or dynamic SLA is required and the need for intensive renegotiation of SLAs is very likely 14. To make the system scalable, the WSAgreement standard, and maybe others such as WS-Notification, would have to be extended to allow LMs to re-negotiate the parts of the agreement which do not belong to them (that is, SLAs which LMs are not involved in creating) but which indirectly relate to the SLA that does belong to them. For example, two LMs can have separate agreements with the same or different FW Exec, each concerning a separate portion of the same workflow. In this case, the changes and failures of one SLA severely affect the success of another SLA. Dr Yarmolenko’s preliminary research on the solution for this scenario suggests the involvement of notification protocols (relates to N in the figure) between various parties that share the responsibility for execution of the entire workflow (point 4) and direct re-negotiations (relates to R in the figure) of SLAs between the parties that did not participate in the creation of this SLA initially (points 2 and 3). The stress on LMs and FW Execs can be significantly reduced by carefully designed dynamic agreements (point 1) which agreement terms may refer to as the terms of other agreements (point 2). This means that WS-Agreement, WS-Notification and possibly other standards (for example, involving authorisation) need to be reviewed. Outcome Dr Yarmolenko suggested a use case in which a number of standards need to be reconsidered or extended to support workflow executions. He also suggested several extensions, in particular, the ability of a WS-Agreement document to refer to another WS-Agreement document. He promoted the usage of dynamic SLAs 15 in the context of workflow management and execution; and raised an important question about the readiness and applicability of current standards to the rising interest of users of and experts in the workflows on Grid. The discussion at the meetings initiated a debate which is still ongoing (via email and other 12 CoreGrid EU Network of Excellence, http://www.coregrid.net 13 Rizos Sakellariou and Viktor Yarmolenko, “On the flexibility of WS-Agreement for job submission”, Proceedings of the 3rd International Workshop on Middleware for Grid Computing MGC '05, Grenoble, France, vol. 117, 1-6 (November 2005) 14 Viktor Yarmolenko, Rizos Sakellariou, “Towards increased expressiveness in Service Level Agreements”, Concurrency and Computation: Practice and Experience, vol.19, 1975-1990 (2007) 15 Viktor Yarmolenko, "Dynamic WS-Agreements" Talk at GGF17 , Slides (PDF) Page 173 3 GridNet2 Reports 3.41 Mr Stefan Zasada – UCL telecommunication). It was anticipated that a more developed form of this debate, along with more refined arguments for a number of extensions (mentioned in this report), would be shared at the upcoming OGF meetings to the wider group by Dr Yarmolenko and the initial participants of this discussion to their respective working groups. Dr Yarmolenko planned to put this forward at the upcoming GRAAP-WG meetings, as well as meetings that discuss WS-Notification, reliability and authentication topics. The main participants in the discussions were: • Viktor Yarmolenko, the University of Manchester (GRAAP-WG) • Philipp Weider, University of Dortmund (contributor to JSDL, GRAAP-WG, chair of GSA-RG) • Alexander Papaspyrou, University of Dortmund (co-chair of OGSA-RSS, participant of JSDL, GSA, GSM) • Michael Parkin, Barcelona Supercomputing Center (GRAAP-WG) • Ramin Yahyapour, University of Dortmund (Coordinator of OGF Compute area, participant in GSA-RG, GRAAP-WG, OGSA-RSS-WG and JSDL-WG) • Oliver Waldrich, Fraunhofer-SCAI (Contributor to GSA-RG, GRAAP-WG) • Wolfgang Ziegler, Fraunhofer-SCAI (GRAAP-WG) • Dirk Neumann, Information Systems, Albert-Ludwigs-Universität, Freiburg • Nikolay Borissov, Universität Karlsruhe 3.41 Mr Stefan Zasada – UCL Mr Stefan Zasada is a Grid middleware PhD student at UCL. For a summary of Mr Zasada’s GridNet2 activities, see section 2.41, Mr Stefan Zasada – UCL. 3.41.1 OGF20 OGF20 was the first OGF meeting that Mr Zasada had attended, and in addition to participating in the two groups identified in his initial application (OGSA BES and GIN), GridNet2 funding allowed him to get a feeling for the other activities in the wider OGF. To this end, Mr Zasada attended the following sessions at OGF20: • SAGA • JSDL (Sessions 1 and 2) • GIN (Sessions 1 and 2) • Workflow Discussion • OGSA-BES • OGSA-EMS Page 174 • Vendor and Developer Adoptions • EGEE Applications • OGSA-RSS The groups that were identified in his initial GridNet2 application are covered in more detail below. OGSA-BES Mr Zasada’s involvement included attending the single BES session at OGF 20, taking notes and participating in the discussion. Around the time of OGF20, the BES standard proposal went into its public comment phase. Mr Zasada used this opportunity to comment on the BES specification, based on his experience of using similar services such as GridSAM, and the requirements of the projects in which he is involved. The comments provided sparked several threads of discussion on the OGSABES mailing list. Mr Zasada’s team plan to extend their Application Hosting Environment software, for which he is the development lead, to support generic BES job submission, and the team was in discussion with OMII UK regarding further support for this work. AHE was expected to be a major component in several large-scale international projects (including the EU funded VPH, ContraCancrum, and ImmunoGrid projects) which have requirements to use several different middleware stacks (such as gLite, Unicore and Globus). Job submission to BES interfaces in each of these middleware stacks is essential. As a direct result of GridNet2, Mr Zasada was in a better position to do this, and as a BES client developer, planned to feed back his team’s experiences to the OGSA-BES-WG. (Awareness of what is happening in the JSDL WG, gained in part from attendance at OGF20, was also helping this work). GIN Mr Zasada’s involvement in the GIN-CG included attending both GIN sessions at OGF 20, taking notes and participating in the discussion. Users within his research group, the Centre for Computational Science at UCL, were involved in using resources from several different Grids on a daily basis (including UK NGS, TeraGrid and DEISA), and so they had a deep vested interest in Grid interoperation. GridNet2 funding allowed Mr Zasada to participate in GIN discussions at OGF 20, and also to report back on the work of GIN to users back at UCL. At the time, a lot of the GIN activity was between resource providers. Although no opportunity arose to contribute to GIN activity directly at OGF20, Mr Zasada actively monitored discussions on the GIN mailing lists and the material generated by the GIN group, and planned to contribute wherever possible in the future, should the opportunity present itself. In summary, GridNet2 funding allowed Mr Zasada to participate in the sessions that are of most interest to his user community, for example, contributing comments to the BES standard based on their experience and requirements, and also to feed back information on what the OGF is doing to his community of computational scientists at UCL. 3 GridNet2 Reports 3.42 Dr Ning Zhang – Manchester 3.42 Dr Ning Zhang – Manchester Dr Ning Zhang is a senior lecturer on Distributed Systems Security at the School of Computer Science, University of Manchester. For a summary of Dr Zhang’s GridNet2 activities, see section 2.42, Dr Ning Zhang – Manchester. 3.42.1 OGF19 Dr Zhang attended OGF19 to chair the LoA BoF session there. Dr Zhang initiated and chaired the LoA BoF at OGF 19. Prior to the session, Dr Zhang wrote and disseminated a document, E-Infrastructure Security: An Investigation of Authentication Levels of Assurance (LoAs), which is now available at: http://www.ogf.org/OGF19/materials/561/OGFLoA BoF.pdf The LoA BoF followed the theme of this document, starting with an overview of existing LoA definitions, including those defined by the US government’s Office of Management and Budget (OMB) and the NIST (US National Institute of Standard and Technology). The session then covered the motivations for further work and emphasised that the existing definitions only cater for human-to-machine authentication scenarios. Finally, there was a discussion on how to proceed to address the gaps when current definitions were being applied in Grid/e-Science contexts. • Federated Identity workshop • OGSA Authz WG meeting 3.42.2 OGF20 In her capacity as co-chair of the LoA RG formed at OGF19, Dr Zhang attended OGF20. LOA-RG The newly-formed LOA-RG group met in the first day, and was chaired by Dr Zhang. The meeting was very productive, as the slides of the meeting show. The slides are available at: http://www.ogf.org/OGF20/materials/723/OGF20LoA-RG.ppt The LOA-RG session discussed LoA attributes, credential issuing and security methods. Current and envisaged authentication architectures and infrastructures were suggested, and some weaknesses identified. In particular, four authentication use case scenarios were discussed, namely: 1. End entity to service direct authentication using end-entity credentials 2. End entity to service authentication using proxy credentials stored locally 3. End entity to service authentication using proxy credentials stored remotely 4. End entity to IdP authentication + IdP to service assertion There were 18 attendees at the LoA BoF, and there were a lot of discussions going on. There was a general feeling that the four levels, as specified by the US OMB/NIST, are good ones, but more work and community consensus were required to address the gaps as introduced by the Grid use case scenarios. These gaps covered several aspects, namely: The audience commented on and gave feedback to the presentation, and the group concluded that there were a lot of gaps between the NIST definitions and the LoA use cases identified. The audience was also invited to take part in a LoA survey, part of the activities to understand current LoA-related procedures, processes and practices as taken by IdPs and service providers. The survey is available at: 1. LoA attributes and factors that have not been addressed by the existing standards http://www.es-loa.org 2. Algorithms for the calculation of overall LoA when an authentication process involves a chain of authentication related activities 3. How LoA attribute values may be conveyed from IdP to service providers At the end of the 90 minutes session, it was agreed that the LoA is a very important aspect of Grid security, and a special research group in OGF, the LOA-RG, should be formed to research and understand the issues and gaps in relation to existing LoA definitions and how LoA may be applied in the Grid environment. It was also agreed to add a work item to the proposed OGSA-AuthN-WG to cover the transporting of LoAs between IdPs and service providers, and to leave the definition of the various LoA levels to IGTF. Dr Zhang accepted leadership of this new RG, together with Yoshio Tanaka. In addition to chairing the LoA BoF session, Dr Zhang also attended the following security related meetings: • SAGA security discussions The LOA-RG would play an important role in achieving risk-based authorisation which enables fine-grained and better security protections to Grid resources, especially when more diversified resources are being incorporated into the Grid environment. OGSA Express Authentication Security Profile Dr Zhang also attended the meeting of the express authentication security profile from the OGSA WG. The profiles were very interesting and useful to help with the LOA-RG’s understanding of the Grid users’ authentication requirements, which was an important pre-requisite for the work being carried out by the LOA-RG. The profiles could also shed some light on the identification of LoA attributes and factors in e-Science and Grid contexts. Funding from the GridNet2 was essential to Dr Zhang’s participation and contribution to OGF LOARG. The results from their survey have reinforced Zhang’s view that LoA linked risk-based authorisation is important and attractive to service providers offering resources with varying levels of sensitivity, and/or wishing to tailor their security protections based upon risk levels. Page 175 A GridNet2 Financial Statements A GridNet2 Financial Statements Table 46: Summary of Costs for GridNet2 EPSRC Component Authorised (£) 29,613.00 28,763.49 Travel and Subsistence 243,104.00 261,311.59 Exceptional Items 125,204.00 78,053.95 0.00 620.40 397,921.00 368,749.43 13,622.00 13,231.21 411,542.00 381,980.64 Staff Equipment Sub-total Indirect Costs Total A.1 Expenditure (£) Financial Awards and Claims Table 47: GridNet2 Awards and Claims Name Applied Awarded Claimed Remainder Brunel Mr Xiaoyu Chen £1,000 £1,000 £973.61 £26.39 Dr Akram Khan £1,000 £1,000 £0.00 £1,000.00 Mr Andrew Harrison £4,000 £3,500 £3,451.75 £48.25 Mr Ian Kelley £6,000 £6,000 £5,978.07 £21.93 Dr Omer Rana £9,080 £7,800 £7,375.97 £424.03 Dr Matthew Shields £2,500 £2,500 £2,304.43 £195.57 Dr Ian Taylor £4,500 £4,500 £4,496.30 £3.70 Mr Chris Higgins £23,400 £12,000 £10,604.01 £1,395.99 Dr Amrey Krause £15,000 £8,000 £6,497.63 £1,502.37 Mr Tom Sugden £15,000 £3000 £0.00 £3,000.00 £13,000 £13,000 £12,993.41 £6.59 Cardiff Edinburgh Glasgow Dr Colin Perkins Page 176 A GridNet2 Financial Statements Name Applied Awarded Claimed Remainder Imperial College London Dr Stephen McGough £69,725 £55,800 £52,911.29 £2,888.71 £12,600 £12,600 £12,805.11 -£205.11 £5,000 £5,000 £4,703.35 £296.65 £11,000 £6,000 £5,844.05 £155.95 Dr Mike Jones £8,050 £8,050 £7,624.92 £425.08 Dr Mark McKeown £4,620 £4,620 £0.00 £4,620 Dr Viktor Yarmolenko £1,690 £1,000 £993.44 £6.56 Dr Ning Zhang £4,320 £2,000 £2,000 £0.00 Kent at Canterbury Prof David Chadwick Manchester Mr Donal Fellows Dr Richard HughesJones National e-Science Centre Prof Malcolm Atkinson £30,000 £30,000 £31,892.59 -£1,892.59 Prof Richard Sinnott £30,000 £16,000 £14,106.08 £1,893.92 £9,120 £9,120 £6,902.99 £2,217.01 £2,200 £1,750 £1,639.57 £110.43 £17,623 £17,623 £13,295.77 £4,327.23 Dr Andrew Martin £15,000 £10,000 £4,356.09 £5,643.91 Dr David Wallom £6,300 £6,300 £2,824.88 £3,475.12 £584 £584 £399.99 £184.01 £18,100.36 £11,399.64 National Grid Service Dr Stephen Pickles Newcastle upon Tyne Dr Savas Parastatidis OMII-UK Dr Steven Newhouse Oxford Reading Prof Mark Baker Science and Technology Funding Council Dr Steven Fisher £34,500 £29,500 Page 177 A GridNet2 Financial Statements Name Applied Dr Jens Jensen Mr David Spence Awarded Claimed Remainder £1,614 £1,614 £1,354.20 £259.80 £2,000 £2,000 £1,627.13 £372.87 University College London Dr Shantenu Jha £12,000 £10,000 £10,000 £0.00 Mr Stefan Zasada £5,000 £3,000 £540.30 £2,459.70 £1,500 £1,500 £1,500 £0.00 £378,926 £296,361 £250,097.29 £46,263.71 Westminster Dr Thierry Delaitre Totals Table 48: Summary of GridNet2 Awards by Institution Institution No. of Awards No. of People Allocated Claimed Brunel 2 2 £2,000 £973.61 Cardiff 9 5 £26,080 £23,606.52 Edinburgh 3 3 £23,000 £17,101.64 Glasgow 3 5 £13,000 £12,993.41 Imperial College London 1 3 £55,800 £52,911.29 Kent at Canterbury 1 1 £12,600 £12,805.11 Manchester 6 6 £26,670 £21,165.76 National e-Science Centre 1 2 £46,000 £45,998.67 National Grid Service 1 1 £9,120 £6,902.99 Newcastle upon Tyne 1 1 £1,750 £1,639.57 OMII-UK 1 4 £17,623 £13,295.77 Oxford 2 3 £16,300 £7,180.97 Reading 1 1 £584 £399.99 STFC-RAL 4 5 £33,114 £21,081.69 University College London 2 2 £13,000 £10,540.30 Westminster 1 1 £1,500 £1,500.00 Page 178 B Publications B Publications This appendix lists the publications that have been drafted or published during GridNet2 where GridNet2 members are named as authors or contributors. B.1 Publications with GridNet2 Authors GridNet2 members were named authors for 39 documents. Note Many of the documents listed in Table 49 have multiple authors. Table 49 lists only the GridNet2-funded authors for the documents. Table 49: GridNet2 Publications - Authors Publication Author Group/Area Ref./Other 2nd International Workshop on Campus and Community Grids David Wallom PGS-RG GFD.126 A Collection of Use Cases for a Simple API for Grid Applications Shantenu Jha SAGA-RG GFD.70 A Requirements Analysis for a Simple API for Grid Applications Shantenu Jha SAGA-RG GFD.71 A Simple API for Grid Applications (SAGA) Shantenu Jha SAGA-COREWG GFD.90 Adding Support to XACML for Dynamic Delegation of Authority in Multiple Domains David Chadwick 10th IFIP TC6 TC11 Conf. Advanced Security for Virtual Organizations: The Pros and Cons of Centralized vs Decentralized Security Models David Chadwick CCGrid 2008 Conf. Advanced Security Infrastructures for Grid Education David Chadwick WMSCI 2006 Conf. Richard Sinnott Anthony Stell Richard Sinnott Anthony Stell Attributes used in OGSI Authorization David Chadwick OGSAAUTHZ-WG GFD.57 Authorisation using Attributes from Multiple Authorities David Chadwick WETICE 2006 Best Paper Award Building a Modular Authorization Infrastructure David Chadwick AHM 2006 Conf. Grid Certificate Profile Jens Jensen CAOPS-WG GFD.125 Mike Jones Page 179 B Publications Publication Author Group/Area Ref./Other Grid Economy Use Cases William Lee GESA-WG GFD.60 GHPN-RG GFD.122 Steven Newhouse Grid Network Services Use Cases from the e-Science Community Richard HughesJones Mark Leese Information Dissemination in the Grid Environment – Base Specifications Steve Fisher INFOD-WG GFD.110 Job Submission Description Language (JSDL) Specification v1.0 Donal Fellows JSDL-WG GFD.56 Multiplexing RTP Data and Control Packets on a Single Port Colin Perkins AVT WG (IETF) Internet Draft Obligation for Role Based Access Control David Chadwick SSNDS07 Conf. OGSA® Basic Execution Service v1.0 William Lee OGSA-BESWG GFD.108 Stephen McGough Steven Newhouse Stephen Pickles OGSA® Data Architecture Dave Berry OGSA-D-WG GFD.121 OGSA® EMS Architecture Scenarios, v1.0 Steven Newhouse OGSA-WG GFD.106 OGSA® Resource Selection Services: Specification Donal Fellows OGSA-RSSWG Draft OGSI Authorization Requirements David Chadwick OGSA-AuthzWG GFD.67 Policy for Supporting Grid Education and Training Malcolm Atkinson ET-CG Draft Providing Secure Coordinated Access to Grid Services David Chadwick MGC 2006 Conf. RTP and the Datagram Congestion Control Protocol Colin Perkins AVT WG (IETF) Internet Draft SDP: Session Description Protocol Colin Perkins AVT WG (IETF) RFC4566 Supporting Decentralized, Security focused Dynamic Virtual Organizations across the Grid David Chadwick Conf. Richard Sinnott 2nd IEEE Conf on e-Science and Grid The Open Grid Services Architecture, version 1.5 Dave Berry OGSA-WG GFD.80 The Storage Resource Manager Interface Specification Version 2.2 Jens Jensen GSM-WG GFD.129 Usage Scenarios for a Grid Resource Allocation Agreement Protocol Stephen Pickles GRAAP-WG Draft Page 180 B Publications Publication Author Group/Area Ref./Other Use-Cases for Grid Checkpoint and Recovery Stephen Pickles GRIDCPR-WG GFD.92 Use Cases for Trusted Computing Andrew Martin TCG-RG Draft Andrew Cooper Use of SAML for OGSI Authorization David Chadwick OGSA-AuthzWG GFD.66 Use of SAML to Retrieve Authorization Credentials David Chadwick OGSA-AuthzWG Draft Use of XAMCL Request Context to Obtain an Authorisation Decision David Chadwick OGSA-AuthzWG Draft Web Services Data Access and Integration The Core (WS-DAI) Specification, v1.0 Amrey Krause DAIS-WG GFD.74 Web Services Data Access and Integration The Relational Realisation (WS-DAIR) Specification, v1.0 Amrey Krause DAIS-WG GFD.76 Web Services Data Access and Integration The XML Realization (WS-DAIX) Specification, v1.0 Amrey Krause DAIS-WG GFD.75 Workshop on Grid Applications: From Early Adopters to Mainstream Users David Wallom PGS-RG GFD.68 B.2 Publications with GridNet2 Contributors GridNet2 members were named in acknowledgements or as contributors to 20 documents. Table 50: GridNet2 Publications - Contributors Publication Contributor Group Ref. A Requirements Analysis for a Simple API for Grid Applications Steven Newhouse SAGA-RG GFD.71 A Simple API for Grid Applications (SAGA) Steven Newhouse SAGA-COREWG GFD.90 Stephen Pickles ByteIO OGSA® WSRF Basic Profile Rendering 1.0 Neil Chue Hong BYTEIO-WG GFD.88 ByteIO Specification 1.0 Neil Chue Hong BYTEIO-WG GFD.87 Defining the Grid: A Roadmap for OGSA® Standards v1.1 Dave Berry OGSA-WG GFD.123 Abdeslem Djaoui Steven Newhouse Page 181 B Publications Publication Contributor Group Ref. HPC Basic Profile, v1.0 Steven Newhouse OGSA-HPCPWG GD.114 Stephen Pickles JSDL HPC Profile Application Extension, v1.0 Donal Fellows JSDL-WG GFD.111 JSDL SPMD Application Extension, v1.0 Donal Fellows JSDL-WG GFD.115 William Lee Stephen McGough Stephen Pickles OGSA-ByteIO Interoperability Testing Specification Neil Chue Hong OGSA-ByteIOWG Draft OGSA® Basic Security Profile 1.0 – Core Dave Berry OGSA-WG GFD.86 OGSA-WG GFD.106 OGSA-WG GFD.120 OGSA-WG GFD.59 Abdeslem Djaoui Steven Newhouse OGSA® EMS Architecture Scenarios, Version 1.0 Dave Berry Donal Fellows Stephen McGough OGSA® Glossary of Terms v1.6 Dave Berry Abdeslem Djaoui Donal Fellows OGSA Profile Definition v1.0 Dave Berry Abdeslem Djaoui Donal Fellows Stephen McGough Steven Newhouse OGSA® Resource Selection Services: Specification Stephen Pickles OGSA-RSS-WG Draft OGSA® Security Profile 1.0 – Secure Channel Dave Berry OGSA-WG GFD.99 OGSA-WG GFD.72 Abdeslem Djaoui Steven Newhouse OGSA® WSRF Basic Profile 1.0 Dave Berry Abdeslem Djaoui Steven Newhouse Policy for Supporting Grid Education and Training Mark Baker ET-CG Draft SAGA Extension: Service Discovery API Steve Fisher SAGA-WG Draft Usage Record – Format Recommendation Donal Fellows UR-WG GFD.98 Page 182 B Publications Publication Contributor Group Ref. Web Services Agreement Specification (WSAgreement) Donal Fellows GRAAP-WG GFD.107 Steven Newhouse Stephen Pickles Page 183 C Standardisation Participation by Area C Standardisation Participation by Area The following table lists the areas of standardisation in which GridNet2 members were involved. GridNet2 members were involved in 49 areas. Table 51: Standardisation Participation by UK e-Science Staff Group People Applications and Programming Models Environment Life Sciences Grid RG Oluwafemi Ajayi, NeSC Richard Sinnott, NeSC Ian Taylor, Cardiff Production Grid Management RG David Wallom, Oxford Simple API for Grid Applications RG Mario Antonioletti, Edinburgh Steven Fisher, STFC-RAL Neil Chue Hong, OMII-UK Shantenu Jha, UCL Mike Jones, Manchester Amrey Krause, Edinburgh Stephen McGough, Imperial Vesselin Novov, Imperial Steven Newhouse, OMII-UK Stephen Pickles, NGS Ian Taylor, Cardiff Elias Theocharopoulos, Edinburgh Stefan Zasada, UCL Architecture OGSA Basic Execution Service WG Malcolm Atkinson, NeSC William Lee, Imperial Stephen McGough, Imperial Steven Newhouse, OMII-UK Savas Parastatidis, Newcastle upon Tyne Stephen Pickles, NGS Stefan Zasada, UCL Page 184 C Standardisation Participation by Area Group People Open Grid Services Architecture WG Dave Berry, NeSC Abdeslem Djaoui, STFC-RAL Donal Fellows, Manchester William Lee, Imperial Stephen McGough, Imperial Steven Newhouse, NGS Savas Parastatidis, Newcastle upon Tyne Omer Rana, Cardiff Ian Taylor, Cardiff Ning Zhang, Manchester Reference Model WG Donal Fellows, Manchester Amrey Krause, Edinburgh Semantic Grid RG Omer Rana, Cardiff Compute Grid Resource Allocation Agreement Protocol WG Oluwafemi Ajayi, NeSC Donal Fellows, Manchester William Lee, Imperial Stephen McGough, Imperial Stephen Pickles, NGS Omer Rana, Cardiff Viktor Yarmolenko, Manchester Grid Scheduling Architecture RG Donal Fellows William Lee, Imperial Stephen McGough, Imperial High Performance Computing Profile WG Malcolm Atkinson, NeSC Donal Fellows, Manchester, Stephen McGough, Imperial Vesselin Novov, Imperial Stephen Pickles, NGS OGSA Resource Selection Services WG Donal Fellows, Manchester Stephen McGough, Imperial Stephen Pickles, NGS Stefan Zasada, UCL Page 185 C Standardisation Participation by Area Group People Data Data and Access Integration Services WG Mario Antonioletti, Edinburgh Malcolm Atkinson, NeSC Dave Berry, NeSC Savas Parastatidis, Newcastle upon Tyne Neil Chue Hong, OMII-UK Amrey Krause, Edinburgh Elias Theocharopoulos, Edinburgh Data Format Description Language WG Mario Antonioletti, Edinburgh Malcolm Atkinson, NeSC Amrey Krause, Edinburgh GridFTP WG Malcolm Atkinson, NeSC Grid File System WG Malcolm Atkinson, NeSC Dave Berry, NeSC Neil Chue Hong, OMII-UK Mike Jones, Manchester Grid Information Retrieval WG David Wallom, Oxford Grid Storage Management WG Malcolm Atkinson, NeSC Dave Berry, NeSC Jens Jensen, STFC-RAL Mike Jones, Manchester Information Dissemination WG Malcolm Atkinson, NeSC Abdeslem Djaoui, STFC-RAL Steve Fisher, STFC-RAL Network Measurements WG Richard Hughes-Jones, Manchester Mark Leese, STFC-RAL OGSA ByteIO WG Malcolm Atkinson, NeSC Neil Chue Hong, OMII-UK Amrey Krause, Edinburgh Page 186 C Standardisation Participation by Area Group People OGSA-Database Access and Integration WG Oluwafemi Ajayi, NeSC Mario Antonioletti, Edinburgh Malcolm Atkinson, NeSC Neil Chue Hong, OMII-UK Amrey Krause, Edinburgh Savas Parastatidis, Newcastle Richard Sinott, NeSC Anthony Stell, NeSC Elias Theocharopoulos, NeSC OGSA Data WG Malcolm Atkinson, NeSC Dave Berry, NeSC Mario Antonioletti, Edinburgh Neil Chue Hong, OMII-UK Amrey Krause, Edinburgh Elias Theocharopoulos, Edinburgh OGSA Data Movement Interface WG Mario Antonioletti, Edinburgh Dave Berry, NeSC Neil Chue Hong, OMII-UK Amrey Krause, Edinburgh e-Science Astronomy Applications RG David Chadwick, UKC Richard Hughes-Jones, Manchester Steven Newhouse, OMII-UK David Wallom, Oxford Audio/Video Transport WG Colin Perkins, Glasgow Certificate Authority Operations WG Oluwafemi Ajayi, NeSC David Chadwick, UKC Jens Jensen, STFC-RAL Mike Jones, Manchester Richard Sinnott, NeSC Curricula Development Mark Baker, Reading Education and Training CG Malcolm Atkinson, NeSC Donal Fellows, Manchester Jipu Jiang, NeSC Amrey Krause, Edinburgh Anthony Stell, NeSC Page 187 C Standardisation Participation by Area Group People Grid Computing Environments RG Ian Kelley, Cardiff Grid Interoperation Now CG Oluwafemi Ajayi, NeSC Malcolm Atkinson, NeSC Dave Berry, NeSC Thierry Delaitre, Westminster Jens Jensen, STFC-RAL Mike Jones, Manchester William Lee, Imperial Stephen McGough, Imperial Steven Newhouse, OMII-UK Stephen Pickles, NGS Richard Sinnott, NeSC Stefan Zasada, UCL David Wallom, Oxford Humanities, Arts and Social Sciences CG Omer Rana, Cardiff Ian Taylor, Cardiff Multiparty Multimedia Session Control WG Colin Perkins, Glasgow Workflow Management RG Andrew Harrison, Cardiff Ian Kelley, Cardiff Stephen McGough, Imperial Vesselin Novov, Imperial Omer Rana, Cardiff Matthew Shields, Cardiff Richard Sinnott, NeSC Ian Taylor, Cardiff Elias Theocharopoulos, Edinburgh Enterprise Enterprise Grid Requirements RG Oluwafemi Ajayi, NeSC Storage Networking CG Richard Hughes-Jones, Manchester Infrastructure Appliance Aggregation RG Mike Jones, Manchester Omer Rana, Cardiff Ian Taylor, Cardiff Grid and Virtualization WG Page 188 Richard Hughes-Jones, Manchester C Standardisation Participation by Area Group People Grid High-Performance Networking RG Richard Hughes-Jones, Manchester Mark Leese, STFC-RAL Network Mark-up Language WG Richard Hughes-Jones, Manchester Network Measurements WG Richard Hughes-Jones, Manchester Mr Mark Leese, STFC-RAL Scheduling and Resource Management Configuration Description, Deployment and Lifecycle Management WG William Lee, Imperial Stephen McGough, Imperial Stephen Pickles, NGS Grid Laboratory Uniform Environment WG Donal Fellows, Manchester Mike Jones, Manchester Job Submission Development Language WG Donal Fellows, Manchester Mike Jones, Manchester William Lee, Imperial Stephen McGough, Imperial Stephen Pickles, NGS Omer Rana, Cardiff Stefan Zasada, UCL OGSA Resource Usage Service WG Xiaoyu Chen, Brunel Donal Fellows, Manchester Mike Jones, Manchester Stephen McGough, Imperial Vesselin Novov, Imperial Richard Sinnott, NeSC Usage Record WG Xiaoyu Chen, Brunel Donal Fellows, Manchester Mike Jones, Manchester David Wallom, Oxford Security Firewall Issues RG Richard Hughes-Jones, Manchester Mike Jones, Manchester Mark Leese, STFC-RAL Andrew Martin, Oxford David Spence, STFC-RAL Page 189 C Standardisation Participation by Area Group People Levels of Authentication Assurance RG Mike Jones, Manchester David Chadwick, UKC Ning Zhang, Manchester OGSA Authorization WG Oluwafemi Ajayi, NeSC David Chadwick, UKC Jipu Jiang, NeSC Mike Jones, Manchester Andrew Martin, Oxford Richard Sinnott, NeSC David Spence, STFC-RAL Anthony Stell, NeSC Elias Theocharopoulos, Edinburgh Trusted Computing RG Andrew Martin, Oxford David Wallom, Oxford Page 190 D Events D Events GridNet2 members were able to attend 45 separate events, as listed in the following table. Table 52: GridNet2 Events Event Date Location GN2 Attendees ETSI Meeting – Nov 06 30 Nov – 1 Dec 06 Sophia-Antipolis, France 1 GFSG F2F – Jan 06 11-13 Jan 06 San Francisco, US 3 GFSG – Feb 06 11-12 Feb 06 Athens, Greece 1 GFSG – May 06 7-9 May 06 Tokyo, Japan 1 GFSG F2F – Jul 06 26-27 Jul 06 London, UK 1 GFSG – Sep 06 9-10 Sep 06 Washington DC, US 1 GFSG – Dec 06 6-8 Dec 06 Amsterdam, The Netherlands 1 GGF14 26-30 Jun 05 Chicago, US 3 GGF15 3-6 Oct 05 Boston, US 11 GGF16 13-16 Feb 06 Athens, Greece 17 GGF17 10-12 May 06 Tokyo, Japan 10 GGF18 11-14 Sep 06 Washington DC, USA 19 ICEAGE Curricula Development Workshop 14-15 Feb 08 Brussels, Belgium 1 IETF64 6-11 Nov 05 Vancouver, Canada 1 IETF65 19-24 Mar 06 Dallas, US 1 IETF66 10-14 Jul 06 Montreal, Canada 1 IETF67 5-10 Nov 06 San Diego, US 1 IETF68 18-23 Mar 07 Prague, Czech Republic 1 IETF69 21-28 Jul 07 Chicago, US 1 IETF70 2-7 Dec 07 Vancouver, Canada 1 INFOD-WG F2F – May 06 14-18 May 06 New York, US 1 INFOD-WG F2F – Feb 07 3-8 Feb 07 Knoxville, US 1 ISSGC 07 8-20 Jul MarieFred, Sweden 1 Page 191 D Events Event Date Location GN2 Attendees LGC Workshop 2-3 May 06 Castleton, UK 1 OGC Committee Meeting Jun 06 Edinburgh, UK 1 OGF19 29 Jan-2 Feb 07 Chapel Hill, US 10 OGF20 7-11 May 07 Manchester, UK 23 OGF21 15-19 Oct 07 Seattle 19 OGF22 25-28 Feb 08 Cambridge, US 16 OGF GIN F2F – Mar 07 14 Mar 07 CERN, Switzerland 1 OGSA F2F 12 15-19 Aug 05 Sunnyvale, US 1 OGSA F2F 13 16-20 Jan 06 Sunnyvale, US 3 OGSA F2F 14 4-7 Apr 06 Sunnyvale, US 1 OGSA F2F 15 17-20 Jul 06 Argonne, US 1 OGSA F2F 16 8-17 Nov 06 Tampa, US 1 OGSA F2F 17 14-16 Mar 07 Belmont, US 1 OGSA F2F 18 13-15 Aug 07 Sunnyvale, US 3 OGSA F2F 19 9-11 Jan 08 London, UK 2 SAGA Design and Implementation Meeting 05 Jan 06 - 1 SLA F2F – Mar 08 11-14 Mar 08 Karlsruhe and Dortmund, Germany 1 SUMOVER – Nov 05 28-30 Nov 05 London, UK 1 SUMOVER – Apr 07 15 Jan 07 London, UK 1 SuperComputing 06 8-17 Nov 06 Tampa, US 1 SuperComputing 07 10-16 Nov 07 Reno, US 2 TC-WG Members Meeting 14-16 Nob 06 Texas, US 1 Page 192 D Events D.1 GridNet2 e-Science Workshop (OGF21) Prof Omer Rana, Dr Stephen McGough and Dr Ian Taylor organised the GridNet2 e-Science workshop that took place at OGF21. The aim of the workshop was to disseminate information about the work being done by GridNet2 members to both the GridNet2 members and the broader Grid community who attend OGF meetings, and to encourage communication between the different areas of OGF in which GridNet2 members participate. For full details about the workshop, see: http://wiki.cs.cf.ac.uk/twiki/bin/view/Sandbox/OpenGridForum21 The GridNet2 e-Science Workshop was held over three sessions at OGF21, and in total, 19 people attended the three sessions. Each of the three sessions had a theme: • Use of novel computer science techniques • Demonstration provided through applications of e-Science concepts • Standards activities and coordination across standards bodies • The following people gave presentations at the GridNet2 e-Science workshop: Session 1 • Prof Rana, Dr Taylor and Dr McGough introduced the workshop. • Dr Shields gave a presentation in the area of workflow on the work being done at Cardiff on workflow sharing and the work of the WFM-RG. There was also input from Mr Kelley, Mr Harrison and Dr Taylor. • Mr Higgins gave a presentation about the geo-spatial work in which he is involved. • Mr Fellows gave a presentation about OGSA, JSDL, Usage Record, Resource Selection and the Reference Model. • Prof Sinnott gave a presentation about security. • Dave De Roure, who was not funded by GridNet2, gave a presentation about semantic Grids, and e-Arts and e-Humanities. Session 2 • Prof Chadwick gave a presentation about the work of the OGSA-AuthzWG over the last four years. • Dr Jha gave a presentation about SAGA. • Prof Rana gave a presentation about the work of the GRAAP-WG. • Dr Hughes-Jones gave a presentation about the NM-WG and also the work of the GFSG The presentations were followed by a brief discussion. Session 3 Dr McGough gave a presentation about JSDL, QoS, workflow, and OGSA. Session 3 concluded with a discussion about the sustainability of the work being done and the way forward for such work. A description of the points raised in the discussion is available at: Page 193 D Events http://wiki.cs.cf.ac.uk/twiki/bin/viewfile/Sandbox/OpenGridForum21?rev=2 ;filename=ogf21-workshop.ppt The workshop played a valuable part in opening up the work being done by individuals in the different areas, and the possibilities for continuing this work and disseminating the information in the future. Page 194 Glossary Glossary A A AA Attribute Authority AAA Authentication, Authorisation and Accounting. The principal components of security. An IETF working group A-WARE An easy Way to Access Grid Resources Axis An open source, XML-based Web service framework Abilene The Internet2 network for the US research and education community. B AD Area Director B2B Business-to-business AFP Application Filter Profile – an HPCBP extension BeinGrid Business Experiments In Grid AG Access Grid BeSTMan Berkely Storage Management System BIRN Biomedical Informatics Research Network BLO Business Level Object An advanced videoconferencing application that combines audio and video tools to allow people in different locations worldwide to meet in a virtual venue. AHM All Hands Meeting BLOBs Binary Large Objects AIDS Application Instance Document Schema BoF Birds-of-a-feather meeting BPEL AIST National Institute of Advanced Industrial Science and Technology (Japan) Business Process Execution Language BPMN Business Process Modelling Notation BREIN Business objective-driven Reliable and Intelligent Grids for Real Business BRIDGES Biomedial Research Informatics Delivered by Grid Enabled Services AJAX Asynchronous Javascript and XML AKOGRIMO Access to Knowledge through the Grid in a Mobile World An EU project for distributed computing AMR Adaptive Multi-Rate compression An audio data compression scheme C AMUC Associated Motion Capture User Categories CA Certificate Authority API Application Programming Interface caGrid APP-AGG Appliance Aggregation The service oriented infrastrucure for the Cancer Biomedical Informatics Grid. CAOPS Certificate Authority Operations. An OGF research group. ARC Advance Resource Connector AssessGrid An EU project to develop and integrate methods for risk assessment and management in all Grid layers Astro RG Astronomy Applications Research Group An OGF working group. CardSpace Client software, from Microsoft, for the Identity Metasystem. CAS Community Authorization Service CASTOR CERN Advanced Storage Manager CATNETS A project for the evaluation of the Catallaxy paradigm for decentralised operation of dynamic application networks. CCLRC An OGF research group. AT Application Template – an HPCBP extension Authn Authentication Council for the Central Laboratory of Research Councils. Authz Authorisation Merged with PPARC to form the STFC in April 2007. AVT Audio-Visual Transport CCR Concurrency and Coordination Page 195 Glossary C D Runtime DCCP Datagram Congestion Control Protocol Deegree An open source community providing OGC interfaces. DEGREE Dissemination and Exploitation of Grids in Earth Science DEPR Data End Point Reference An OGF working group DESY Deutsches Elektronen-Synchrotron CG Community Group DFDL Data Format Description Language CIM Common Information Model DGAS Data Grid Accounting System CIS Credential Issuing Service DigSigs Digital signatures CMOS Complementary Metal Oxide Semiconductor Discovery Net CMS Content Management System Condor A scalable, customised job scheduler Software for arranging database access and knowledge discovery procedures, and provides a means of describing workflow between analysis service providers, data owners and the scientists who arrange and execute the workflows CondorShibboleth An integration of Condor and Shibboleth DIT Directory Information Tree (X.500) Contra Cancrum Clinically Oriented Translational Cancer Multilevel Modelling DM Distributed Management DMIS CP Certificate Policy Data Movement Interface Standardisation CPS Certification Practices Statement DMTF Distributed Management Task Force CREAM Computing Resource Execution and Management DOM Document Object Model An architecture and library for monitoring WS-Agreements Dorian A Grid user management service CREMONA DPM Disk Pool Manager CROWN China Research and development environment Over Wide-area Network DRMAA Distributed Resource Management Application API Credential Validation Service DSLA Dynamic Service Level Agreement CVS An EGEE project to bring together the GMES and Grid communuties DSS Decentralised Software Services Cyclops DTI Department of Trade and Industry (UK), now known as the Department for Business Enterprise and Regulatory Reform CCT Center for Computation and Technology at LSU CDL Configuration Description Language for CDDLM CDDLM Configuration Description, Deployment and Lifecycle Management D D-Grid Deutscher-Grid DTLS Datagram Transport Layer Security DAGMan Directed Acyclic Graph Manager DyVOSE Dynamic Virtual Organisations in e-Science Education DAIS Data Access and Integration Services An OGF working group E DAML DARPA Agent Markup Language EACL Extended Access Control List DART Digital Audio Retrieval using Triana EC-GIN Europe-China Grid Internetworking dCache A system for storing and retrieving large amounts of data, distributed among a large number of heterogeneous server nodes EDA Electronic Design Automation EDG European Data Grid EDGeS Enabling Desktop Grids for e-Science DCC Page 196 Digital Curation Centre Glossary E F An EU project EDINA National Academic Data Centre, based in Edinburgh. EGA Enterprise Grid Alliance Merged with the GGF in 2006 to form the OGF EGEE Enabling Grids for e-Science eMinerals A project to use computer simulations performed on molecular length and time scales to address environmental issues EnLIGHTened Computing A project to develop dynamic, adaptive, coordinated and optimised use of distributed computing resources EoL Encyclopedia of Life EPCC Edinburgh Parallel Computing Centre EPR End Point Reference EPSRC F2F Face-to-face meeting FI Firewall Issues An OGF research group FIPA Foundation for Intelligent Physical Agents An IEEE Computer Society standards organisation that promotes agent-based technology and the interoperability of its standards with other technologies FSP File Staging Profile – an HPCP extension FTS File Transfer Service G G-lambda A project to establish a standard web services interface between Grid resource management systems and network resource management systems. Engineering and Physical Sciences Research Council G-PBox A policy management service for production Grids. eRep An EU project to provide theorydriven and empirically backedup guidelines for designing reputation technologies G-QoSM Grid Quality of Service Management GACL Grid Access Control List ESB Enterprise Service Bus GAUSSIAN Software for computation chemistry ESLEA Exploitations of Switched Lightpaths for e-Science Applications GCE Grid Computing Environments ESLoA e-Infrastructure Security Levels of Assurance GDI-Grid Spatial Data Infrastructure Grid GÉANT ESNet Energy Sciences Network A European multi-gigabit computer network for research and education purposes ESP-Grid A project to investigate how Shibboleth offers solutions for Grid authentication, authorisation and security issues. GEESE Grid Enabling EDINA Services GEMEPS Grid Enabled Microarray Expression Profile Genesi-DR Ground European Network for Earth Science Interoperations – Digital Repositories Genesis II An open source, standards-based Grid platform designed to support both high-throughput computing and secure data sharing GEODE Grid Enabled Occupational Data Environment GEOSS Global Earth Observation System of Systems GFAC Grid Forum Advisory Committee GFS Grid File System ESRC Economic and Social Research Council ESRI Environmental Systems Research Institute ET Education and Training An OGF research group An OGF community group ETF Engineering Task Force eToken A USB-based strong authentication and password management solution ETSI European Telecommunications Standards Institute eVLBI electronic Very Long Baseline Interferometry An OGF working group GFSG Grid Forum Steering Group Page 197 Glossary G GGF G Global Grid Forum GRIDCC Grid Enabled Remote Instrumentation with Distributed Control and Computation GRIDCPR Grid Checkpoint and Recovery Merged with the Enterprise Grid Alliance to form the OGF in 2006 GGUS Global Grid User Support An OGF working group Part of EGEE GHI Genetic Healthcare Initiative, now known as GSSFHS GHPN Grid High Performance Networking GridFTP A high-performance, secure, reliable data transfer protocol GridPP UK Grid for Particle Physics Research GRIDREL Grid Reliability and Robustness An OGF research group GI Geographic Information GID Group identifier GIN Grid Interoperation Now An OGF community group GIN-DATA The GIN-CG’s interoperation project for data management and movement GIN-JOBS The GIN-CG’s interoperation project for job description and submission GIR Grid Information Retrieval An OGF working group An OGF research group GRIDRPC Grid Remote Procedure Call An OGF working group GridRUS An implementation of the OGF Resource Usage Service that provides a queryable store of usage records for Grid infrastructure providers. GridSAM A web service for job submission and monitoring that uses the JSDL specification GridShib An integration of Shibboleth and Globus GridSite A set of extensions to the Apache web server and a toolkit for Grid credentials, GACLs and HTTP(S) protocol operations GIS Geographic Information Systems GLASS GLASgow implementation of Shibboleth GLIF Global Lambda Integrated Facility GridSphere gLite Next generation middleware for Grid computing An open source, portlet-based Web portal GRIDVIRT Grid and Virtualisation Globus An open source, software toolkit used for building Grid systems and applications GLOW Grid Laboratory of Wisconsin GLUE Grid Laboratory Uniform Environment An OGF working group Grouper An open source toolkit for managing group information across integrated applications and repositories GSA Grid Scheduling Architecture An OGF research group An OGF working group GLUEMan A framework to manage information providers for GLUE 2.0 GME Grid Management Entity GMES Global Monitoring for Environment and Security GNAB2 GridNet2 Advisory Board GNI Grid Network Interface GRAAP Grid Resource Allocation Agreement Protocol An OGF working group GridHypervisor Page 198 The prior name for OpenNebula, a virtual infrastructure engine GSI Grid Security Infrastructure GSM Grid Storage Management An OGF working group gSOAP A cross-platform development toolkit for C and C++ SOAP XML Web services GSSFHS Generation Scotland, Scottish Family Health Study GT4 Globus Toolkit version 4 GTS Grid Trust Service GUMS Grid User Management System GUNI Global University Network for Innovation Glossary G GUS I with Grid technologies Grid User Services An OGF research group H HDTV High Definition Television HEP High Energy Physics HLRSStuttgart INFN Istituto Nazionale di Fisica Nucleare INFOD Information Dissemination An OGF working group INSPIRE Infrastructure for Spatial Information in Europe High Performance Computing (Höchstleistungsrechenzentrum) at the University of Stuttgart IPR Intellectual Property Rights IPTV Internet Protocol Television HPC High Performance Computing iRODS i Rule Oriented Data Systems HPC4U Generic, modular Grid middleware which enables an increased level of Fault Tolerance and covers multiple administrative domains ISI Institute of Information Science at the University of California ISSGC International Summer School on Grid Computing HPCBP High Performance Computing Basic Profile ITU HPCP High Performance Computing Profile International Telecommunication Union ITU-T The ITU Telecommunication Standardisation Sector IVOA International Virtual Observatory Alliance An OGF working group HPDC High Performance Distributed Computing HSM Hardware Signing Model J JBI Java Business Integration Identity and Access Management JDBC Java Database Connectivity Part of MAMS JISC IANA Internet Assigned Numbers Authority Joint Information Systems Committee JRA Joint Research Activity ICE Internet Connectivity Establishment JSDL Job Submission Description Language ICEAGE International Collaboration to Extend and Advance Grid Education JSON JavaScript Object Notation JXTA Juxtapose I IAM ICENI Imperial College e-Science Networked Infrastructure ID Information Document An OGF term. IDL Interface Description Language IdP Identity Provider IETF Internet Engineering Task Force IEEE Institute of Electrical and Electronics Engineers IGTF Internation Grid Trust Federation iLBC Internet Low Bit Rate Codec IMAA Institute of Methodologies for Environmental Analysis ImmunoGrid A computer model of the human immune system, implemented An open source, peer-to-peer networking protocol, originated by Sun Microsystems K Kepler An open source, scientific workflow system Kerberos A network authentication protocol Knoogle An OMII-UK project to develop a framework that can be used for creating and deploying bespoke and reconfigurable matchmaking and brokerage components. KnowARC Grid-enabled Know-how Sharing Technology Based on ARC Services and Open Standards KPI Key Performance Indicator Page 199 Glossary K Kryptus M Broker A Brazilian HSM project myVocs L A virtual organisation collaboration system LCAS Local Centre Authorisation Service N LCG LHC Computing Grid nanoCMOS LCMAPS Local Credential Mapping Service A project to investigate the challenges in nanoCMOS design. LDAP Lightweight Directory Access Protocol nanoHUB A web-based resource for research, education and collaboration in nanotechnology LFC LCG File Catalogue NARA LGC Lightweight Grid Computing National Archives and Records Administration (US) LHC Large Hadron Collider NaReGi National Research Grid Initiative (Japan) LIBI Laboratorio Internazionale di Bioinformatica (Italy) NAT Network Address Translation LoA Levels of Authentication Assurance NCeSS National Centre for e-Social Science (UK) NCSA National Center for SuperComputing Applications (US) NDL Network Description Language NEESgrid Software that links earthquake researchers across the U.S. with computing resources and research equipment NERC Natural Environment Research Council (UK) NERSC National Energy Research Scientific Computing Center (US) An OGF research group LSF Load Sharing Facility Job scheduler software from Platform LSG Life Sciences Grid An OGF research group M MAMS Meta Access Management System MDS Monitoring and Discovery Service MIB Management Information Base NeSC National e-Science Centre (UK) MIDCOM Middlebox Communication NextGRID A project to enable a an architecture for Next Generation Grids which will enable their widespread use by research, industry and the ordinary citizen An IETF working group NGS National Grid Service (UK) An array of tools and middleware components to share and manage data and metadata in Grid and distributed computing environments NHS National Health Service (UK) NICE A provider of software products aimed at company-wide management and optimisation of computing resources NIST National Institute of Standards and Technology (US) NM Network Measurements An IETF working group MMUSIC Mobius Multiparty Multimedia Session Control MoU Memorandum of Understanding MPI Message Passing Interface MSI Minority Serving Institutions my Experiment A virtual research environment that allows the sharing of digital items associated with research, particularly scientific workflows. myProxy mySRB Page 200 An OGF working group. NML Network Markup Language An OGF working group. Open source software for managing security credentials for the X.509 public key infrastructure NoE Network of Excellence NomCom Nominations Committee (OGF) A web-based browse and search interface to the Storage Resource NorduGrid A Grid research and development collaboration by Nordic academic Glossary N O and research institutes for the ARC middleware ns-2 A discrete event simulator for networking research NSF National Science Foundation (US) NTT Nippon Telegraph and Telephone Corp. O OASIS Organization for the Advancement of Structured Information Standards OCSP Online Certificate Status Protocol OGC Open GeoSpatial Consortium OGC Collision A project to make geospatial data available securely on the UK Grid, and increase the availability of Grid resources, collaboratively, with UK e-infrastructure organisations OGF technological infrastructure for the rapid prototyping and development of knowledge-intensive, distributed, open services for the Semantic Grid OpenID An open and decentralised identity system, independent of any corporation ORION Ontario Research and Innovation Optical Network OSD Object based Storage Device OSG Open Science Grid OWL Web Ontology Language P P-Grade A portal that provides access to the services supplied by multiple Grids. P2P Peer-to-Peer P2PS Peer-to-Peer Simplified Open Grid Forum PA Preparation Agreement Created from the merger of the GGF and the EGA in 2006. PBS Service orientated architecture, Grid infrastructure software for cluster and Grid computing PE Preservation Environments OGSA Open Grid Services Architecture OGSA Authz OGSA Authorisation OGSA-BES OGSA Basic Execution Services Pegasus A Grid workflow system An OGF working group. PEP-CVS OGSA interface for reading input and output from a stream of bytes Policy End Point – Credential Validation Service PEP-PDP Policy End Point – Policy Decision Point perfSONAR Performance Service Oriented Network Monitoring Architecture PERMIS An infrastructure that provides management of privileges and authorisation policies and allows applications to make authorisation decisions. PGP Pretty Good Privacy PGPGrid Pepper’s Ghost Productions Grid OGSAByteIO An OGF research group An OGF working group An OGF working group. OGSA-D OGSA Data An OGF working group. OGSA-DAI OGSA-Database Access and Integration OGSA-DMI OGSA Data Movement Interface An OGF working group. OGSA-EMS OGSA Execution Management Services OGSA-RSS OGSA Resource Selection Services An EPCC project that explores the use of the Grid for the production of computer-generated animations An OGF working group OGSI Open Grid Services Infrastructure OIL Ontology Inference Language OMB Office of Management and Budget (US) OMG Object Management Group OMII Open Middleware Infrastructure Institute (UK and Europe) OntoGrid A project to produce the PGS Production Grid Services An OGF research group PIPSS PPARC Industrial Programme Support Scheme A knowledge transfer scheme that supports the development of collaborations between UK Universities, CERN, ESO, ESA, UK industry and research sector organisations Page 201 Glossary P R Protocol PhEDEx Physics Experiment Data Export PI Principal Investigator RTP Real-time Transport Protocol PKI Public Key Infrastructure RTSP Real-time Streaming Protocol PMA Policy Management Authority RUS Resource Usage Service POSIX Portable Operating System Interface – an IEEE standard family for the API, shell and utilities interfaces for UNIX-compatible software PPARC Particle Physics and Astronomy Research Council Merged with CCLRC to form the STFC in April 2007 An OGF working group. S S-OGSA Semantic Open Grid Services Architecture S/MIME Secure/Multipurpose Internet Mail Extensions SAGA A Simple API for Grid Applications PRACK Provisional Response Acknowledgement SAML Security Assertion Markup Language PSNC Poznan Supercomputing and Network Centre SARoNGS Shibboleth and Grid integration project Pubcookie A standalone login server and modules for common web server platforms allowing single sign-on authentication for websites in an institution. Q QoS Quality of Service R A follow-on to ShibGrid and SHEBANGS SAW-GEO Semantically-Aware Workflow Engines for Geospatial Web Service Orchestration SBRN Scottish Bioninformatics Research Network SCUFL Simple Conceptual Unified Flow Language SDI Spatial Data Infrastructure SDO Standards Development Organisation RA Registration Authority RAL Rutherford Appleton Laboratory rat Robust Audio Tool RG Research Group RDF Resource Description Framework RDF-S RDF Schema RealityGrid A computing infrastructure for supporting the modelling of complex, condensed matter systems. SEE-GEO Secure Access to Geospatial Services SeRQL Sesame RDF Query Language REST Representational State Transfer Sesame RESTful A system which uses the principles of REST An open source framework for the storage, inferencing and querying of RDF data RFID Radio Frequency Identification SGAS SweGrid Accounting System RM Reference Model SGE Sun Grid Engine OGF and IETF are SDOs. SDP Session Description Protocol SDT Service Description Terms SecPAL Security Policy Assertion Language A MicroSoft project. An open source, batch-queueing system An OGF working group RMS Resource Management System RNS Resource Namespace Service RPC Remote Procedure Call RTCP Real-time Transport Control Page 202 SHEBANGS Shibboleth Enabled Bridge to Access the National Grid Shibboleth A standards-based, open source, software package for web single sign-on across or within organisational boundaries Glossary S S A prototype system that allows NGS users to access NGS facilities securely through federated authentication mechanisms employed at their institutions STS Security Token Service SUMOVER Support for Mbone Videoconferencing for the Research community Signet A privilege management toolkit SVC Scalable Video Coding SIMDAT An EU project to create data Grids for process and product development, using numerical simulation and knowledge discovery SweGrid Sweden’s national computational resource SWITCH Swiss Academic and Research Network. ShibGrid SIP Session Initiation Protocol SLA Service Level Agreement SLO Service Level Objective SN Storage Networking An OGF community group. SNIA Storage Networking Industry Association SOA Service Oriented Architecture SOAP A protocol for exchanging XMlbased messages SOCKS A security package that allows a host behind a firewall to access resources outside the firewall SOKU Service Oriented Knowledge Utilities SORMA T TAGPMA The Americas Grid Policy Management Authority Taverna An open source workflow system TC Trusted Computing An OGF research group TCP Transmission Control Protocol TeraGrid An open, scientific discovery infrastructure that combines resources at eleven partner sites to create an integrated, persistent computational resource TFRC TCP Friendly Rate Control Theora An open source, video compression format Self Organising ICT Resource Management Torque A project for the the development of methods and tools for an efficient market-based allocation of resources on the Grid A resource manager that provides control over batch jobs and distributed compute nodes Triana An open source, problem solving environment comprised of a visual interface and data analysis tools. TrustCoM An EU project to create an environment for trust, security and contract management in B2B collaborations SP Service Provider SPARQL SPARQL Protocol and RDF Query Language The acronym SPARQL is a recursive acronym. Speex An open source, audio compression format codec SPMD Single Process, Multiple Data SRB Storage Resource Broker SRM Storage Resource Manager SSH Secure Shell StorM A Grid Storage Resource Manger STFC Science and Technology Facilities Council Created from the merger of CCLRC, PPARC and the nuclear physics responsiblities previously managed by EPSRC in April 2007 U UDAP Universal Dynamic Activity Package UDP User Datagram Protocol UID User identifier UML Unified Modeling Language UNICORE Uniform Interface to Computing Resources UNSDI United Nations Spatial Data Infrastructure UPDATE A SIP method that allows a client to update the parameters of a session (such as the set of media streams and their codecs) but has no impact on the state of a dialog Page 203 Glossary U W An OGF research group. UPM Universidad Politécnica de Madrid UR Usage Record WFMC Workflow Management Coalition An OGF research group. WFS Web Feature Service WG Working Group WHIP Workflows Hosted in Portals project WLCG Worldwide LHC Computing Grid WMProxy Workload Management Proxy V Vine VIOLA A modular, extensible Java library that inherits the GridPortlets’ resources model and generalises it. Vertically Integrated Optical testbed for Large Applications A component for accessing the gLite WMS Virtualisation software for running multiple PC-based operating systems simultaneously on one workstation WMS VM Virtual machine WPS Web Processing Service VMWare Proprietary virtualisation software WS Web Services VO Virtual Organisation WS-BPEL Web Services Business Process Execution Language VoIP Voice over IP WS-CNP Web Services Contract Net Protocol VOMRS VOM Registration Service VOMS Virtual Organisation Membership Service WSDAIOnt Web Services Data Access and Integration in OntoGrid WS-DAIR Vorbis An open source, professional, audio encoding and streaming technology Web Services Data Access and Integration – Relational Realisation WS-DAIX VOTES Virtual Organisations for Trials and Epidemiological Studies Web Services Data Access and Integration – XML Realisation WS-GRAM VPH Virtual Physiological Human Web Services Grid Resource Allocation and Management VPMan A project to integrate VOMS and PERMIS for superior secure Grid management. WS-I Web Services Interoperability WS-RF WSRF Web Services Resource Framework WS-X Various Web Services, either finalised or in progress WSN Wireless Sensor Network WSS Web Security Service Virtual PC VPN Virtual Private Network VRVS Virtual Room Videoconferencing System W WAS Web Authentication Service WAYF Where Are You From A way for a Service Provider to find out where to send Shibboleth requests to get you authenticated and make a decision on whether you are allowed to access a resource Workload Management System Web Map Service X X.501 An ITU-T framework for the X.500 series of standards. X.509 An ITU-T standard for public key and privilege management infrastructures. WCS Web Coverage Service XACML WebDAV Web-based Distributed Authoring and Versioning eXtensible Access Control Markup Language XAM eXtensible Access Method Workshop on Enabling Technologies: Infrastructures for Collaborative Enterprises Xen A hypervisor that is an open source industry standard for virtualisation An IEEE event. XFire A next-generation JAVA SOAP stack framework WETICE WFM Page 204 Workflow Management Glossary X XLST XML Stylesheets XPath XML Path Language is a language for selecting nodes from an XML document. In addition, XPath may be used to compute values (strings, numbers, or boolean values) from the content of an XML document. XPDL XML Process Definition Language XQuery An XML query language that is designed to query collections of XML data Z ZRTP An extension to RTP for cryptographic key agreement. Page 205