OGSA Teleconference - 21 July 2004
==================================

* Participants

  Dave Berry (NeSC)
  Ian Foster (ANL)
  Larry Flon (ISI) 
  Bill Horn (IBM)
  Hiro Kishimoto (Fujitsu)
  Andrew Grimshaw (UVa)
  Fred Maciel (Hitachi)
  Andreas Savva (Fujitsu)
  David Snelling (Fujitsu)

  Minutes: Andreas Savva  

* August F2F update

  - Hiro is talking with Heather about having a common meeting
    - Participants in the WSDM F2F must be members of OASIS. So the
      common meeting should be done in the OGSA meeting.
  - The UC-WG F2F is also that week.

* Data F2F update

  A lot of people came to the F2F. There were no Data area directors
  but a couple of other people from related Data WGs (e.g., DAIS)
  attended.

  Started off with idea presentations which showed that there is a lot
  of commonality in ideas. Discussion topics
  - Metadata
  - File management
  - Distribution (global name to file)
  - Replication & Caching (usually for performance but also for
    usability---the aim affects what needs to be done)

  It is important to arrange common meetings with other Data WGs to
  make sure that we are not stepping into other people's territories.

  Andrew described the role of OGSA-WG. The group does architecture
  and spawns WG or collaborates with existing WGs if they want to join
  the work. But OGSA-WG is not constrained by what's there already.
  This role wasn't really known by others at F2F.
  - (DaveB) Andrew explaining that "the architecture must be cohesive"
    might have come across as saying that OGSA-WG might
    ignore/redo/... other work, which is not the intention. It
    understated the negotiation aspect of the OGSA-WG work.
  - (DaveS) There is GFSG permission to overlap efforts with other
          groups. Also there have been at least hints of this
          approach. And getting from design to a more detailed level
          will make it more obvious.
  - (DaveB) INFO-D overlap is the most obvious (and probably also with
          the global filesystem)
          - INFO-D overlap is not just with data work but also with
            logging, etc.
          - One suggection from the INFO-D chair was to pause their
            discussion until higher level work is done
  - (DaveS) Or maybe it should be the other way. It is not a good idea
            to pause an active WG.
  - (DaveB) Their concern is how to best use their time. 
  - Could also spend time on ogsa-wg data calls as well.
  - There is a need to figure out where the overlap is first

  ACTION: Andrew and Dave S to get something written down on the
          'process' discussed above. At the moment it is only in the
          GFSG minutes, which are not public.

** OGSA V2 Outline proposal
        
  - One other thing that came up from the Data F2F is the V2 outline
    proposal. It is uploaded to gridforge.

  - Level 1: Putting service descriptions and scenarios in seems to
    make it too long. The doc should be of a size that people can read
    easily and get an idea of overall architecture.

  - Level 2: Each design team does their service descriptions and
    scenarios in different documents. 
    - Expected level of description: about an english paragraph or two
      on how services interact + semantics.

  - Level 3: Detailed service specs and wsdl as separate
    documents. These are recommendation documents by each WG.

  One concern is that Level 2 might also be better done by a WG not by
  an OGSA-WG design team. On the other hand it is the sort of thing
  that is needed as the first step for starting up a WG. It might be
  more concrete than what some WGs actually have at the moment.

  (DaveS) If the end result is something that can be implemented and
          tested and therefore needs to be shown to interoperate then
          it should be done in a WG.
 
  - By (close-but-not-quite-right-)analogy to DMTF, Level 2 is UML
    (detailed but not normative), Level 3 is MOF (normative).

  - The discussion and acceptance of this V2 outline should be an
    agenda item for the F2F.

  - Another issue was the nature of the documents (information vs
    recommendation).
  - Also whether a design team can actually produce a document or
    not. And it if does is it a recommendation or information
    document. The idea at the moment is that design teams only do
    focus work, no formal documents.

** Context based document

  It is a strawman (Level 2) for one of the identified services.

** Design Philosophy memo

  - A previous version was sent out at the May F2F
  - There have been some additions as people reviewed it
  - Not sure if it should be in the OGSA specification but it would be
    nice to have somewhere

  ACTION: Andrew to fill in gaps (push people who have volunteered to
          provide text) and produce a new version.

** Other issues from Data F2F

  - Naming
    - Resolution
  - Factory
  - Migration vs data movement
  - change management

* EMS update

  - Need to review the DMTF work to identify where there are overlaps
  - CIM Job seems to be different from what is described as a job in
    OGSA.
  - Some concerns were raised that the CIM level of abstraction seems
    to be lower than what expected. It is exposing a lot of details
    that would be assumed hidden for a lot of the EMS scenarios.

  - What is the connection between CIM and 
    - SRIM? It is a group working in CIM
    - WS-CIM? It is CIM and web services. A mapping of existing CIM
      model to Web Services. They are discussing their charter at the
      moment. The relation of WS-CIM to WSDM is still not clear.
              
  - (Andrew) Requirements are more important than implementation at
    the moment.

  - Terminology: model (defined in CIM), the rendering of the model
    (e.g., by WS-CIM) and CIMOM is the implementation

* Naming proposal in OGSA version 1, draft 19

  - Text on p54 of v1 captures the naming proposal well
  - (Andrew) Resource handles are important. There is also a potential
    requirement from the Data group that they should be comparable.
  - There are lots of different naming schemes