eScience meets eFrameworks 28th April 2006 NeSC, Edinburgh A PPARC funded project Tony Linde Programme Manager AstroGrid overview Virtual Observatory infrastructure “The VO vision can be summed up as the desire to make all archives speak the same language − all searchable and analysable by the same tools, all data sources accessible through a common interface, all data held in distributed databases that appear as one.” Andy Lawrence, 09/2003 AstroGrid approach 28-Apr-2006 Build infrastructure first VObs Web Mix-and-match, plug-and-play, … eScience meets eFrameworks Tony Linde 2 AstroGrid project Duration: Funding: Personnel: Goal: Scope: AG1 Phase A: AG1 Phase B: AstroGrid-2: 28-Apr-2006 Sept 2001 – Dec 2007 £7.7M (PPARC) ~26 (23.4 FTE) Develop VObs Infrastructure Deploy UK VObs Astrophysics, Solar, STP, … Optical, X-Ray, Radio, … Sept 2001 – Dec 2002 Analysis, R&D, Architecture Jan 2003 – Dec 2004 Build, test & deliver V1.0 Jan 2005 – Dec 2007 Develop & deploy eScience meets eFrameworks Tony Linde 3 AstroGrid consortium Edinburgh Leicester Cambridge MSSL JBO RAL QUB From 2005, AstroGrid-2 + 28-Apr-2006 Bristol Exeter Leeds Portsmouth eScience meets eFrameworks Tony Linde 4 Wider involvement Europe Euro-VO: http://euro-vo.org VOTech project: http://www.eurovotech.org FP6 funded: €6M: 2005-7 Follows on from AVO (FP5 funded) International IVOA: http://ivoa.net Closer collaborations RVO JVO SAAO NVO 28-Apr-2006 eScience meets eFrameworks Tony Linde 5 Project Approach Agile: iterative & incremental development Open project: Static Wiki Forum Jabber http://www.astrogrid.org http://wiki.astrogrid.org http://forum.astrogrid.org via jabber.astrogrid.org Open Source code Academic Free License AstroGrid releases 28-Apr-2006 http://www.opensource.org/licenses/academic.php http://software.astrogrid.org eScience meets eFrameworks Tony Linde 6 AstroGrid infrastructure 28-Apr-2006 eScience meets eFrameworks Tony Linde 7 Deployments 1 28-Apr-2006 eScience meets eFrameworks Tony Linde 8 Deployments 2 28-Apr-2006 eScience meets eFrameworks Tony Linde 9 Deployments 3 28-Apr-2006 eScience meets eFrameworks Tony Linde 10 How do the VO components that implement these interfaces relate to each other? 28-Apr-2006 eScience meets eFrameworks Tony Linde 11 DSA: Architecture ADQL SkyNode SIAP/ SSAP Web Svc Translator D S A Plugins: JDBC/Db FITS etc VOTable (CEA) 28-Apr-2006 eScience meets eFrameworks Tony Linde 12 Registry We can access data but how do we know where it can be found? Knowledge of the location and nature of resources is critical for all this to work The AstroGrid Infrastructure supplies a Registry (IVOA compliant of course!) which contains this information Registry is searchable by humans and by machines. 28-Apr-2006 eScience meets eFrameworks Tony Linde 13 Common Execution Architecture Why would we want CEA/UWS? Same reason we need universal access to data AstroGrid supports the concept of workflow Essentially, remote execution of special scripts or jobs Allows data access and processing to be undertaken at source or where the compute power lies The Grid concept! The AG Infrastructure helps make the details transparent, keeping focus on the science. 28-Apr-2006 eScience meets eFrameworks Tony Linde 14 Workflow 28-Apr-2006 eScience meets eFrameworks Tony Linde 15 Compute grid Common Execution Connector Application as .exe file (local) Common Execution Connector Application as .exe file (grid) Common Execution Connector Application as .HTTP service Common Execution Connector Application as Java class JES Registry 28-Apr-2006 eScience meets eFrameworks Tony Linde 16 MySpace If all this happens remotely, where do the results end up? Distributed data access and processing requires distributed data storage MySpace is the AstroGrid solution VOSpace/VOStore under review at IVOA Distributed, virtual data storage that appears to the user as a homogeneous collection of files Database tables coming soon! 28-Apr-2006 eScience meets eFrameworks Tony Linde 17 Distributed Storage DSA Process Process Desktop MySpace 28-Apr-2006 eScience meets eFrameworks Tony Linde 18 MySpace 28-Apr-2006 eScience meets eFrameworks Tony Linde 19 AstroGrid Actors Astronomers Incl PIs, management structure, … Instrument operators, … Data curators Publishers Administrators Systems managers Developers Standards definers … 28-Apr-2006 eScience meets eFrameworks Tony Linde 20 AstroGrid and eFrameworks: Issues Security Shibboleth authentication: Single-signon Authorisation Common policy requirements? Service discovery Resource registries Need for commonality? Will others need to get at astro resources? 28-Apr-2006 Common identifiers? eScience meets eFrameworks Tony Linde 21 AstroGrid and eFrameworks: Other Issues Service execution Workflow Shared storage Accounting Naming! e.g. VO Bottom line 28-Apr-2006 Standards & interoperability How do we get W3C, GGF, IVOA, IE & project standards to work? eScience meets eFrameworks Tony Linde 22