Leicester: WPA1/WPA0: Architecture Focus Meeting
Draft Minutes 3 March 2002 by NicholasWalton
Present: Nic Walton (NAW), Tony Linde (TL), Guy Rixon (GTR), Clive
Page (CP), Bob Mann (BM), John Sherman (JCS) (after 14.20)
Time: 13.00-17.00
1. Talking through the agenda as posted on the
wiki page. TL noted
that the layer diagram was only in a draft form and was open to
amendment.
TL noted that ACTIONS, except where specified, are for completion
by the end of Q3 (e.g. 31 May 2002).
2. TL indicated that Q3 work would still be largely design and
research work. GTR stressed that any prototypes developed in this
period will not be final, and therefore the interfaces are liable
to change.
3. Science and User cases:
WPA1. Abstraction of science and user cases: process is to scope the
current full set in the VO wiki, and determine which are relevant to
the priority data sets for the 3 yr project. these then become the
Astrogrid set of cases.
ACTION NAW by 20020322
4. WPA1. Activity diagrams (= flow chart) - NAW to talk to the person
responsible for that case area - talk through it. NAW and that WP
manager would together generate the relevant flow diagram. (NAW needs to make
sure that there is sufficient text info associated with each
science/user case.) NAW needs to identify high priority data sets for
AstroGrid in Phase-B, and note where they are held (or distributed
from).
ACTION NAW by 20020322
5. WPA2. In order to specify the client manager we need to know how
data is specified and accessed. Anita Richards will be asked to look into
this. Starting point is CP's paper (shortly to be posted on the
Wiki) on existing VO prototypes.
ACTION TL by 20020320: ask Anita Richards to generate this report
Object models, component and implementation diagrams (OCI) for Level
3. GTR - need to see a web service prototype with Globus
installed. (relevant for tools and functions services in L2).
ACTION GTR by 20020531: Prototype a computation on one or more grid
nodes (e.g. some Iraf tasks). Deploy and run a computation intensive
process on the grid.
6. VO interface - GTR needs to look into whether or not WSDL is
appropriate at the higher levels.
GTR is going to rewrite the layer diagram (check Wiki for updates) -
refactoring the top part of the diagram to break down what is in the
browser and what is in the server.
ACTION on ALL - draw up a plan of what they are doing and planning
for Q3 - quarterly forecasts by March 8, March 6 for the reports.
Part of WPA2: look at where jobs will be run (job optimisation
services in the VO services block). whole issue of estimating run
times, does the user want to proceed with the job - idea of
monitoring jobs.
JCS: there is a worry about that are the boundary conditions on the
architecture. We need to be using existing technologies where
possible. CP/TL noted that we were endeavouring to do just this.
JCS: when do we talk to the data centres about they will interface
with the
AstroGrid system?
It was agreed that it was not sensible to discuss this with the data
centres as a whole until Phase-B of
AstroGrid was better
specified. Thus TL would re-visit this question in Q4. TL/NW did
note though that many of the archive centres that
AstroGrid would
interact with in Phase B were in fact already members of the
AstroGrid consortium.
CP noted that
AstroGrid could offer useful capabilities to data
centres in the area of authorisation/ certification.
GTR/JCS; a design review of the architecture will be needed to
ensure that it will meet the demands of the use cases which are
defining it. it was AGREED that this will happen at the next
AstroGrid Collaboration meeting that should be held in June 2002.
ACTION TL: fix dates and set up a Collaboration meeting for the
week starting 24 June 2002.
7. WPA4 - CP wondered how does this work fit into other work being
done outside of
AstroGrid, or in partnership with
AstroGrid:
e.g. Interoperability work withing Opticon/AVO.
NW asked what is the status of the VOTable proposal. Does WSDL
supersede it? No, VOTable holds the XML descriptors.
ACTION CP: Before generating an OCI for data access, WPA4 will make
a proposal for the data query language which will be exposed by the
data access web services. (The discussion focused on VOTable and
xquery.)
Part of WPA4 will include the development of a runtime estimator,
perhaps located in Level 1 of the Layer diagram.
BM noted that SX issues run time estimates - but it only knows
about the time needed to complete a query on the database. This
doesn't take account of distributed databases, speed of networks
etc.
8. WPA9: draft designs for Level 4: client and browser based UI. JCS
noted that this could use the WPA5 pilot matching optical
and X-ray data as setting the requirements for a first pass
prototype.
WPA9 will assess how runtime estimates/feedback from runtime
monitoring etc will be integrated into the UI.
9. Research:
ACTION JCS: WPA3: assess how the proposed OGSA initiative can be
integrated into the
AstroGrid architecture.
ACTION CP: WPA4: assess Grid initiatives (e.g. DBTF) and their
likely impact in the area of data mining.
10. Technical approach for product development would be addressed by
the Technical working group.
11. TL, through WPA0 would specify
AstroGrid Phase-B (by end Q3)
12. Prototypes to be developed in Q3
- JCS: WPA9: user interface, browser and client
- GTR: WPA2: certification, authentication and authorisation system
- GTR: WPA2: limited resource directory (the two WPA2 prototypes might be combined into one).
- Elizabeth Alden: WPA2/5: a web service interface to access a data set
- GTR: deploy and run a computational module on a grid machine (see 5.)
13. There was some discussion as to the main risks associated with Q3
work. The risk, the probability of occurrence, and its impact if it
did occur were noted. Where appropriate a workaround was
identified.
| Q3 Risks | Probability | Impact on Q4 |
| 13.1 Will OGSA be sufficiently defined in Q3? | med | low |
| (note: our OCI diagrams may turnout to be wrong if we define them as OGSA services which get radically altered from the present description) |
| 13.2 Staff effort resourcing: not all in place by 1 May | high | med |
| (workaround - we plan not to use new effort in Q3) |
| 13.3 AGLI not agreeing phase-B plan | low | med |
| 13.4 Use cases not fully developed by the end of March to a level that they drive all functionality that might be required by AstroGrid. | med | high |
| (workaround - include features for development in Phase-B that are not explicitly required by the use cases: although these would presumably have a use case derived for them) |
| 13.5 The prototypes are all not developed on time in Q3 | med | med |
| (workaround - make sure at least two get done) |
| 13.6 Research does not produce useful results | low | low |
| 13.7 Prototypes and research lead to incompatible results | low | low |
| (workaround: June architecture review to prioritise) |
| 13.8 Lack of cooperation with other bodies (eg AVO, CDS) | low | low |
| 13.9 Output from WP not usable for input to architecture | low | high |
| 13.10 AstroGrid staff effort not working to the agreed time scales | low | med |
| (workaround: PM will act when alerted) |
| 13.11 Collab tools integrity: backup of AstroGrid project collaboration servers | low | high |
| (workaround: CP to look into backup policy) |
Meeting closed at 17.00
--
NicholasWalton - 02 Mar 2002
The actions from his meeting should be taken to include those listed in the
agenda, particularly the OCI diagramming, unless specifically revoked above.
--
TonyLinde - 18 Mar 2002