GGF10 - OGSA-WG Session 1
=========================

* OGSA Evolving, Jeff Nick

  WSRF & OGSI

        WSRF and OGSI was covered in more detail by Steve Tuecke's
        talk in a previous session. Focus mainly on what transition
        means to OGSA.

        OGSI introduction was a bold move. It attempted to address
        Grid requirements but was not well aligned with the direction
        Web Services were taking. It was not well integrated with WS
        standards and was seen as fracturing the community. (E.g., the
        industry has a lot of investment in tooling; OGSI, though not
        wrong, did not fit well (different paradigm) with current
        tooling.)  WSRF move to address this and other problems.

        Other concerns addressed:
          - OGSI had too much functionality in one spec. To encourage
            adoption it was split up into a number of (composable)
            specs.

            One of them (WS-N) is still too big. Refined further into
            3 smaller specs.

          - Too object oriented: Separate GS into service and state
            (resource).
        
        WSRF refactoring and alignment led to huge commercial support
        for WSRF. Shows acceptance of OGSI idea; success moving
        forward.

        Note that OGSI functionality is preserved in WSRF. GT4 to show
        that evolution is possible. Higher level services will not be
        affected by move to WSRF.

  OGSA connection

        A big WSRF motivation is systems management. Need some basic
        common functionality to support broader functions relating to
        resource management (talking about resources and managing
        them) etc; manageable resources.

        OGSA to standardize on functions too; standardize on more than
        syntax.

        Cross-working with other standards bodies, e.g., DMTF on
        manageable resources and models, OASIS for basic web services
        functionality, manageability.

        OGSA work to fill in the boxes (functions) by looking at use
        cases and coming up with concrete functionality that solves
        hard problems.

        Job scheduling as an example of a discipline that is well
        understood and where GGF has lots ot expertise. GGF can lead
        this area.

  Questions

    Connection of WSRF and WS-Architecture?
        WS-A is W3C's attempt to draw distinction from low level and
        high level functionality thereby identifying where it fits in
        (in other words it is an articulation of policy).

        (Low level functionality (wire level specs) done in W3C while
        higher level stuff (WSRF, WS-N) in OASIS or GGF.)

        Is there any conflict? 
           There might be tension but not conflict.
        WS-A doesn't seem to be driving WS process. 
   

    Working on use cases seems to be a good idea. In particular has
    there been work on security and requirements to solve basic 
    problems?

        There is ongoing work within the group to identify the gap
        between WS security standards and OGSA.
        

* Progress report and future plans, Hiro Kishimoto

  Activity summary 
     - 2 interims since last GGF, 2 calls a week, active gridforge
       updates
     - Active participation from both academia and industry
 
     - First use case draft sent to GGF Editor.
     - OGSA specification updated to include some of the latest work.

    Architecture (big picture) in OGSA; detailed work in
    expert/specialized WGs. Collaboration and joint work/sessions.

    Deliverables: 
      - Use Cases (final draft v1 GGF10)
      - OGSA spec. (final draft GGF11)
      - Roadmap (first draft at GGF11)

  Questions

    Not technology for technology's sake. What is industry's viewpoint
    on OGSA; where is it pushed, and which products are going to use
    it?
        Business Grid Project, Globus, IBM

    There is talk of OGSA being built around WSRF. What happens if
    WSRF does not materialize by next year? (Or if industry support is
    not there?)

      - From the point of view of the spec, the WG has already
        discussed this issue (Sep'03) and decided that OGSA is/should
        be architecturally independent of the base WS specs used to do
        the rendering to the resources. Currently expressing the OGSA
        spec in OGSI; will probably express in WSRF.

      - When we(implementers) come round to building things
        will have to make the hard decisions.

      - Standards take time to evolve. 
        - IBM intends to build on WSRF and OGSA.

    Is it correct to do WSRF rework in ogsa after GGF10?
   
      Current draft is using OGSI terminology. The rework addresses
      terminology/language. It does not imply (or force) an
      implementation.

    How about other specs from other authors, e.g., WS-Eventing?
  
      WS-Eventing does not seem to be very far from what is in the
      WS-N.

    Revisiting question of WSRF support from other vendors.

      WSRF was just announced. TC to be formed in OASIS. Other
      organizations can sign up. Until the press release cannot really
      speak about lack of support from other vendors. Initial
      indications show a lot of interest.

      Also there are different types of vendors; some vendors such as
      those in system management might adopt the WSRF specs earlier
      than others.