Minutes of the OGSA Data telcon, 8th June 2005

0) Actions arising

- Dave to reconfigure tracker mail
- Dave to reply to Torsten of the TM-RG, inviting him to our session
- Ann will organise the security notes we've taken so far into a coherent memo that we can
take to the security working groups.
- Dave will discuss security with the NeSC security people.
- Dave will approach the OGSA WG security design team about a joint meeting.
- Allen will approach the OGSA-AuthZ WG about a joint meeting.
- Stephen to iterate on the replication use case.
- Dave to revamp section 3 of the architecture document.

1) Early discussion

Roll call 
  Dave Berry, NeSC
  Mark Morgan, University of Virginia
  Stephen Davey, NeSC
  Allen Luniewski, IBM
  Fred Maciel, Hitachi
  Ann Chervenak, ISI

Apologies
  Mario Antonioletti, EPCC

The minutes were approved and the agenda was accepted.


2) Action report

- Dave to approach transaction RG re GGF session. [See below]
- Peter to write a first cut at the Storage section of the architecture
document [Done]
- Dave to try to arrange data transfer discussion  [Scheduled for GGF]
- Allen to cut and paste material into document before F2F [Done]
- Dave to forward final F2F meeting info [Done]
- Dave to mail round draft F2F agenda [Done]


3) WG status 

- Trackers
Dave set up a tracker for the architecture document.  He attempted to have this
send notifications to the mailing list but this did not work.

- GGF working drafts
The architecture document has been submitted.  We agreed to keep this version
unmodified until after GGF.  Proposed changes can be circulated as memos.

- GGF session plan
We have requested 3 sessions, both by mail and by tracker.  We haven't heard
anything back and we're getting nervous.

Torsten has replied about the joint session we proposed with the TM-RG.  The
TM-RG are wrapping up their activity and will present their results at GGF14.
We agreed to invite him to our session as well, so that we could have a working
session that concentrated on our requirements.


4) Security requirements for data grids

Some conversation has taken place by e-mail.

Ann said that there are multiple models and that we need to lay out the space
rather than prescribe a particular solution.  E.g. Bill and Peter have outlined
different approaches to security. 

We are not sure of the status of GGF security WGs.  There is consultation within the Globus and EGEE projects but how does this relate to standards?

We don't have a complete understanding of authorisation in the data world.  Globus uses two levels of authorisation - the VO level allows access from a VO point of view but this may be overridden by the resource owner.  Capabilities can be as fine-grained as you like for VO authorisation but scalability is a limit on this.

Replica case - users might be able to see replicas at one site but not others, by access control of the storage managers.  Another case is that a replica might be a superset of the original, combining multiple requests for a replica, then the authorisation at the replica end might be more complex than just copying the original data.

Ann will organise the notes we've taken so far into a coherent memo that we can
take to the security working groups.

Dave will discuss this with NeSC security people.
Dave will approach the OGSA WG security design team about a joint meeting.
Allen will approach the OGSA-AuthZ WG about a joint meeting.


5) Overall architecture

Following London F2F, Stephen has looked at use case templates with the aim of writing
up a data replication scenario.  He decided that some of the sections from the OGSA
use case  templates might not be appropriate for a scenario.

Dave offerred to expand the description of the OGSA framework in section 3 of the document.  This will position the data services within the overall OGSA work. E.g. naming, security, scheduling, advance reservation.

OGSA scheduling would like our suggestions about which properties scheduling will need, e.g. logical locality.  We're not sure that this is the best approach - scheduling will likely require negotiation plus it also needs input from network services.  Perhaps a memo outlining data considerations for security would be better.


6) Wrap up

DONM: Next week, possibly reviewing memos written as part of actions.