OGSA Teleconference April 05, 2004
==================================

* Participants

  Ian Foster (ANL)
  Andrew Grimshaw (UVa)
  Bill Horn (IBM)
  Hiro Kishimoto (Fujitsu)
  Fred Maciel (Hitachi)
  Andreas Savva (Fujitsu)
  Latha Srinivasan (HP)
  Ravi Subramaniam (Intel)
  Jem Treadwell (HP)
  
  Minutes: Andreas Savva

* Early discussion

** Teleconference minutes April 1 - No comments, approved.

** Agenda bashing

*** Infrastructure (in)dependence discussion (review)
    
    - Last call we did something like straw poll. 
    - Consistent consensus towards WSRF on calls so far.

    - Comments on Savas' email: 
      - Disagreement that WS-Security is "highly stable".
      - WS-I (& WS-Security) may be too low level for OGSA. The group
        will have to build more if do not adopt WSRF.
        - E.g., we need something like a name which can be covered by
          EPR to some extent.
      - Concern that standardization of WSRF will take time.
        - OGSA spec in June won't be detailed to the level of
          normative WSDL. WSRF concepts will be used for concreteness;
          to make the spec easier to write (and read).
        - Specs spun off from OGSA spec. will have to go to more
          detail (normative wsdl). But by that time we expect that
          WSRF details will be more stable.

   Proposal: Document decision (why) and how we are using WSRF in
             introduction of the OGSA spec.

   - The group will have to be ready to defend this decision at the
     next GGF (and possibly earlier too).

   Proposal: Get a presentation ready addressing all points raised
             so far. Ideally have someone that is not deeply connected
             to WSRF to give the talk; preferably a person from
             industry.

   Action: Write up text explaining the direction the group is taking
           (what we are thinking of doing & why) and post to the
           list. 
           - Ian to write first draft and circulate among small group
             before posting to list: Hiro, Jem, Ravi, Andrew.

   - Ravi's email: Problem space in abstract terms; definition in
     concrete terms. 
     - Abstract definition of the problem might satisfy some people
       pushing for abstract (ogsa) definition.
     - In some sense have been doing this already (e.g., PE).
     - In theory good idea; concern with whether we have enough time.
     - Use spirit of ravi's email to draw up draft.

* OGSA document re-organization discussion

   - See http://www-unix.gridforum.org/mail_archive/ogsa-wg/2004/04/msg00020.html

   - Some discussion on Ravi's pre-ggf10 slides
     (https://forge.gridforum.org/projects/ogsa-wg/document/OGSA_overview_slide.pdf/en/1)
     - Instance diagram   
     - Fabric: things that every service has; how much should go there.
     - Black ovals are VOs
     - Offers a different perspective from layering; using layering
       introduces a number of problems.

   - Do we need to include taxonomy? (ch.4 in Hiro's proposal)
     - It is non overlapping grouping but since Ch.3 discusses how the
       services are used to provide certain capabilities, and Ch.5
       details the services the taxonomy seems redundant.

   Agreed: Remove taxonomy section from Hiro's proposal.

   Open issue: How to use Ravi's picture in the document.

* OGSA glossary review plan

  Action: Hiro to update the teleconference schedule so that Jem can
          schedule appropriate glossary terms discussions.

  Action: Jem to take part in CDDLM discussion (call) on provisioning
          and use it for updating the OGSA glossary

  - Andrew is listed against a lot of the terms in the glossary. Some
    of those terms may be obsolete. 

  Action: Jem and Andrew to talk separately about Andrew's items. Identify
          what is obsolete & delete. Update the rest.

  Approach:       
    - For teleconference discussions impose strict time limit (15
      minutes) so we don't waste people's time.
    - Make sure we only discuss terms that have draft definitions.
    - Glossary should have crisp definitions and reference to
      normative spec that defines the term.

  - Should the glossary be part of the architecture doc or not?
     
  Rejected: Include the glossary in the final version of the OGSA
            spec.
  Agreed: Keep glossary separate but linked with the ogsa document.
          - But make sure they are submitted together as a logical
            unit.
          - The glossary version that corresponds to a given version
            of the OGSA specification should be easily available.

  - Should the glossary be structured alphabetically or not? 
    - Might make more sense to divide into broad sections such as
      Provisioning etc.

  - What to do with old definitions?
  
  Agreed: Keep old definitions and mark as deprecated or obsolete.
          - OGSI terms in particular: reference/contrast/relate with
            WSRF terms.

  Action: Jem to followup privately with all people listed as
          (potential) sources in the glossary.

* CMM-WG plan discussion

  - Detailed plan send to key people
  - Work as OGSA-WG design team; need wider OGSA-WG expertise
  - Teleconference on demand; sometimes have slots on OGSA calls
  - People will be asked to attend as needed
    - So teleconf times will be arranged appropriately
  - In addition post CMM call information to OGSA ML (with distinctive
    header) to reach a wider audience
  - More detailed planning leading to GGF11, including major revision
    of documents.

  - Relation to DMTF? 
    - CIM (information via Andrea Westerinen)
    - Utility Computing WG? (pending)