OGSA-WG Teleconference -- 23 October 2003
=========================================

* Participants

  David Berry (UK NeSC)
  Bill Horn (IBM)
  Hiro Kishimoto (Fujitsu)
  Takuya Mori (NEC)
  Fred Maciel (Hitachi)
  Andreas Savva (Fujitsu)
  Frank Siebenlist (ANL)
  David Snelling (Fujitsu)
  Jeffrin Von-Reich (HP)
  Ravi Subramaniam (Intel)
  Chris Smith (Platform)
  Jem Treadwell (HP)

  Minutes: Andreas Savva

* Last teleconf minutes - approved

* Agenda bashing

  - Jeff is travelling and can't join. Postpone the comments
    discussion.

* Next F2F

  - Note that we are planning 2 F2F meetings. There are 2 pages in the
    spreadsheet Hiro sent out. One for Dec the other for Feb. Also
    note the location preference. Some people indicated that the
    likelihood of availability may vary.

  - Agreed: Spreadsheet should also allow people to indicate
    'likelihood'.

  - Other possible dates discussed:
    - Post SC2003(Arizona) (e.g., 11/21,22)
    - Post GlobusWorld(California) 

  - Only Ian has expressed a strong preference for Chicago. (May not
    be able to attend otherwise.)

  - Proposal: Do a 3 day meeting.
    - But then we'd probable like time in the middle to tackle
      homework.

  - Comments: 
    - Indicating preferences on the spreadsheet does not imply a hard
      commitment --- people's availability may change.
      - Preferred: Can probably make it.
      - Available: May be able to make it but inconvenient.

  - Action: Hiro to re-send spreadsheet and add 'likelihood' as an option.

* GGF9 minutes - approved

  - PDFs are readable (checked by Ravi)
  - Action: Ravi to send his notes from platform session to Andreas. (>Done).

  - Action: Fix AuthZ name in Core session minutes is wrong. (>Done).

* GGF9 cross session summary - continue from GESA

** GESA - Want to be part of OGSA. Exploring relationship with GRAAP
   & JSDL.

   - Dave mentioned Jon MacLaren (NeSC) workshop presentation. A
     summary of the issues covered there (and in other presentations)
     might make a good addition to Scheduler/PE section.
   - Workshop: http://www.nesc.ac.uk/action/esi/contribution.cfm?Title=309

   - Bill: Is GESA in OGSA or on top, e.g., using OGSA?
     - Ravi: 2 perspectives
       - Agency that sells things in marketplace
       - Mechanism built in architecture

   - Agreed: Keep watching. 
   - Action: Decide how they fit with us --- inside as part of
             architecture or on top.
   - Action: DaveB & Dave to make document (summary mentioned above)
             for inclusion to OGSA doc.

   - GESA uses WS-Agreement; WS-Agreement is part of architecture &
     GESA pushes down on their requirements; so they are engaged
     with us.
     - Mismatches: 
       - WS-Agreement model - client initiates and server
         responds. But in GESA the other way round is also possible.
         Resources can bid for work. 
     - Currently GESA is ogsi-compliant; looking for OGSA 'compliance'.

   - Action: Symmetry issue to be put to GRAAP-WG. Owner is probably
             GESA-WG but Dave to monitor.

* Core session
--------------

** OGSA-SEC - Alignment with OGSA doc

   - Are there any holes between grid security and ws-security?
     - Probably, but not clear yet.
   - Is proxy based delegation included in the long term?

   Agreed: Since Frank had to drop out early skip OGSA-Sec and AuthZ
           discussion for next time.

   Agreed: Tell people in advance (on the list) that we'll discuss
           these topics and get them to join discussion.

   Agreed: Keep watching

** AuthZ - OGSI based.

   Agreed: Keep watching

   - Comments: 
     - Is fine grain authorization (sde or port level or operation) in
       OGSI?
       - Not yet; But might have some capability (annotations) in
         WSDL-1.2 to extend operations.
       - SDE: could include info in SDE about visibility, access, etc.

** RUS - Want to be part of OGSA. Relation with UR, GESA.

   - Bill: Parallel (tactical) ogsi-compliance and (strategy)
     ogsa-compliance approach.

   - RUS is user of UR. View UR as more generic than job usage.
   - RUS is intended to be part of OGSA.

** UR - Content definition. No obvious conflict.

   - UR seems to be focused on batch jobs; what about network usage
     etc.

   - Nice xml schema for resource consumption of batch jobs
   - Group considers itself finished.

   - Does it belong in our document?
     - Need standard way to represent resource metrics
     - Put it in as informative reference to normative document.
       See [(*) Other discussion] further down.

   - Proposal: It might be an interesting to formulate it in our
               document as a resource usage event.

   - Agreed: We need something in this area and no group is doing what
             we want.

* Platform session
------------------

** CMM - Most work moved to OASIS-WSDM. Do gap analysis to find new
   niche.

   - CMM was spawned by OGSA.

** P2P - Working on requirements. Feedback to OGSA.

   - Agreed: OGSA should be able to support P2P.
     - So getting requirements is important.
     - Fundamentally OGSI can(should be) able to support this model;
       services are peers.

   - Agreed: Have to be careful that OGSA(?) service definition does
             not assume client/server model.

   - Comments: 
     - P2P is wider than what's in this(OGSA-P2P) group.

   - Agreed: We have to re-visit this topic. Defer more general issues
             (e.g., asymmetry) to another teleconf or to the F2F.

* Still remaining: AA

(*) Other discussion

   - Proposal: The OGSA document should contain no normative
               references to current wgs; so as to allow for change.

   - Proposal: Do two definitions: 
               1) an abstract definition not depending in a normative
                  way to existing stuff.
               2) Follow it up with a concrete definition (rendering
                  of the abstract definition) using current stuff.

   - Action: Do a draft normative rendering of OGSA in terms of
             current standards (~1 page?) >Hiro. (This is not urgent.)