Minutes of the OGSA Data WG telcon, 17th May 2007

1) Early discussion:
 
* Roll call 

Dave Berry, NeSC (Note taker)
Mario Antonioletti, EPCC
Allen Luniewski, IBM

The minutes were approved.
 

2)  Action report
 
* New actions

[Allen] Review the policy subsection in the light of the discussion.

[Dave] Make sure that the architecture document explains that we can handle sets of files.

[Allen] Check the relation between RNS and storage-specific directory management.


* Completed actions

[Dave] Reply to Erwin about the Data Area

[Dave] Put current working versions into the "current" directory on GridForge.

[Dave] Prepare foils for OGF20.


* Ongoing actions

[Dave] mail Simon to clarify the "classify" operation on metadata catalogues.

[Chris] send out second draft of GFS scenario.
 
[Dave]  to create or reuse scenario slides for main OGSA slide set.
 
 
* To be added to the trackers
 
[Mike] Section 12 - Metadata Catalogue and Replication - needs to be worked on
[Mario] Describe the semantics of destruction of services/data in the early section. The service has to define it's semantics.
[Mario] Include a paragraph in the Appendix setting the context and scope of each of the specs.
[Mario] Use the data access patterns to motivate the data access section.
[Mario] In the data access section, the create operation needs to be scoped down.  Maybe break it down into separate operations.
 
[?] The glossary needs to be updated.
[?] The cache section needs to be completed.
 
[?] Appendix 1 listing specs – already done?
[?] Appendix 2 summary of general interfaces – to be completed in final draft?
[?] Appendix 3 mappings to specs – to be completed in final draft?
 
[Dave] to revise the EMS/Data scenarios
 
[?] Add Data Caching scenario
[?] Add Data Warehousing scenario
[Dave] Add Metadata Catalogue scenario
 
[?] Incorporate Avaki and Oracle scenarios as motivation.
 
[Dave] What about asynchronous operations and aborting requests? (Suspend/Resume/Abort Request)


3) Progress report

Nothing to report.


4) OGF20

We held one group report session, which generated some interesting discussion among about 30 people.  We also presented our gap analysis at the data area session.

The overriding message is that we should get stuff into public comment a.s.a.p.

Regan suggested that replication should include some distributed operations, e.g. number of replicas, free space available, ...

On Federation, we should check that the text makes clear that federation is opaque rather than transparent, i.e. that the names available to the client are not necessarily related to the names managed by the components.

Allen to review the policy subsection in the light of the discussion.

Dave to make sure that the architecture document explains that we can handle sets of files.

Allen to check the relation between RNS and storage-specific directory management.

Mario has uploaded his notes on the web site.


5) DONM

25th May 2007.  We shall discuss the minimum set of tasks to complete the Architecture document.