OGSA Interim Meeting (OGSA-16) - Day 1, morning
===============================================

  Date: 9 November 2006 (Thursday), morning

* Participants

  Fabio Benedetti (IBM)
  Michel Drescher (Fujitsu)
  Hiro Kishimoto (Fujitsu)
  Fred Maciel (Hitachi)
  Tom Maguire (EMC)
  Steve McGough (Imperial)
  Mark Morgan (UVa)
  Andreas Savva (Fujitsu)
  Ellen Stokes (IBM)
  
  Phone bridge: 
    Donal Fellows (UoM) (until Security session)
    Chris Kantarjiev (Oracle)
    Jay Unger (IBM)
    David Groep (NIKHEF) (from Security session onwards)

  Minutes: Andreas Savva

* Summary of New Actions

  AI-1109am-a: Ellen and Tom will review the error code/faults issue
                further with a view to put a proposal forward to DMTF.
  
* Welcome

* Review of "Containers in OGSA"

  - Need to apply new OGF template (see Gridforge Editor project)
  - The plan is to publish it as an Information-track document and
    contribute it to the DMTF

  - [Sec.2 paragraph2] Container is general and might need
    sub-classing for more specific container types
    - Is the relationship between this document and BES clear enough?
      It is part of the discussion today.

  - In thinking about how an information model relates to a service,
    should we consider defining a BES Service (sub-classed from
    CIM_Service) where the BES Service contains the remaining BES
    container specific factory operations?  (Refer to new visio
    drawing)

  - Revised model suggests 'moving' operations from Container to
    Service (generic ones) and specializations of Service for more
    specific ones (e.g., BES Container)
    - Container would have only a set of attributes that are generic
      enough (perhaps only a type attribute?) that apply to all
      sub-types.
    - Should Queuing Service and J2EE Hosting Environment be
      considered together as possible sub-types?
      - The interesting point is what resources are behind the queuing
        service, for example, rather than what the container is
        'running on'
    - Need some kind of transformation model to build document set(s)
      for XQuery --- the advertisement model
      - (A related topic is the connection to SML)
      - XQuery itself could be one of those tools but this usage is
        different from the one envisaged for matching
    - Clearly this is still in the initial stages and need more work
      to determine what has to be done.

  - Is an attribute that denotes the environment of container needed;
    such as whether the container is a queuing service, a J2EE hosting
    environment, a host (operating system based), etc?
    - Probably, but still have open issues as above.

  - The current version of the BES spec defines 2 attributes: Common
    Name and Long Name.  These do not have to defined in this
    container model because the Container class inherits Caption and
    Description, respectively, from the ManagedElement class.
    - Agreed no problem with this.

  - Are there any other elements of BES that we think are of a general
    nature and should be added to this Container class definition?
    - not at this point

  - open() and close() as detailed below have no inputs, but do have a
    return code. BES analogous operations have no inputs and no
    outputs.  How do we want to define open and close?
    - There is a difference in thinking about these. In particular
      between the operation failing (which in the view of the BES spec
      writers is not possible) and of the operation succeeding and
      returning some abnormal returns that should be modeled as faults
      instead.
    - There is also a discrepancy in the way CIM models everything as
      error codes.

  AI-1109am-a: Ellen and Tom will review the error code/faults issue
                further with a view to put a proposal forward to DMTF.

  - WS-CIM appears not to be the right XML representation to list
    here---from an OGSA point of view.  From a DMTF/CIM point of view,
    WS-CIM can be algorithmically generated (per DSP0230).  For this
    information document, Ellen recommends that it does not include a
    WS-CIM XML representation. 
    - Agreed to remove the WS-CIM XML section from this document

  - Is a Glossary needed?
    - No need. Checked that the Container definition in the OGSA
      Glossary 1.5 is good and the Glossary is referenced.

  - Ellen will produce a revised version

* Document status review

  - Updates directly in the spreadsheet by Hiro; and uploaded to
    Gridforge.

* Security

  - Review of proposed charter for AuthN
  - Current proposal is to focus on protocol aspects; along the lines
    of the OGSA-AuthZ group
  - Is profile accurate here or not? It might be better called an
    instantiation.
  - Should review of existing work be a target? There is a lot of
    existing work in this area and this may be a long task.
  - Is delegation in scope - yes
  - Jim Basney (NCSA) was named as one person to try to get involved
    in this group.
  - Best Practice and Roadmap are mentioned as deliverables.  Why not
    a core specification as well?
    - Agreed to add a title for a recommendation document.
  - There are many specific problems in moving forward. For example,
    why is this topic important, what is it trying to solve. In
    particular what is the specific threat model that it is addressed,
    what threats can this work mitigate, and so on.
    - Agreed to add as part of Best Practices a section on common
      threats and what are the practices that mitigate them.