National e-Science Centre

advertisement
National
e-Science
Centre
th
First meeting of ATF2
29 August 03, e-Science Institute, Edinburgh
The first meeting of ATF2 was held at the National e-Science Institute, Edinburgh on 29th August
2003. It had Steve Tuecke of Argonne National Lab as an invited guest, and was called at
relatively short notice to take advantage of his visit to the UK.
Malcolm Atkinson introduced the meeting by explaining the role of the revised ATF: To look
ahead to the 3 to 6 year horizon and to develop an architectural roadmap for grid architectures.
He planned to get the ATF2 to work intensively for 6 months to produce a first report, have a
"rest" period of 6 months, and then revise the architecture roadmap in another intensive stint.
We viewed "grid" in this context as the means of providing regular and high-level structures over
very large and dynamically varying inter-organisational distributed computational systems that
may be used to share resources and information, to enable collaboration and to operate
distributed infrastructure. We anticipated that grid boundaries would be hard to define because of
many external interactions, e.g. with sensor networks, pervasive computing and business
processes.
He then asked each member of the ATF2 to identify the major perturbing factors on grid
architectures that they foresaw in the ATF2 timescale. After a preliminary discussion, it was
agreed to return to these issues in more detail at later meetings.
The majority of the meeting was spent discussing a presentation by Steve Tuecke attached.
Each topic in the talk generated much discussion. There was detailed consideration of the
extensive interaction between the OGSA/OGSI effort and the web service standards effort. It was
apparent that very substantial effort had been invested, particularly by Steve Tuecke and Steve
Graham in communicating with WS standardization groups. The situation remained confusing,
partly because the shape of OGSI had evolved when there were very few relevant WS proposals
that had been made public and partly because there are many rival proposals in the WS context.
Steve Tuecke remarked that they would, with hindsight have omitted the G from OGSI and have
had a group of independent but compatable specifications. However, the importance of
stablishing a platform on which OGSA discussions could proceed and which enabled
implementations had been and remained the driving factor. It is important to have a platform that
is stable and agreed so that the higher-level architectural design and implementation issues (e.g.
OGSA) can be explored. The platform should be revisited, once the contemporaneous
technological changes were understood and the requirements and workload produced by OGSA
had been identified.
Malcolm Atkinson thanked Steve Tuecke for a very stimulating presentation and discussion, and
the ATF2 members for coming at short notice. He noted the requirement for a longer lead time in
arranging meetings and their agenda.
The meeting was attended by: Malcolm Atkinson, Dave Berry (invited observer), Howard Chivers,
John Darlington, Wolfgang Emmerich, Peter Lamb (invited observer), Beth Plale (invited
observer) and Steve Tuecke (Guest).
Malcolm Atkinson
National e-Science Centre
Download