OGSA Teleconference - 7 June 2007 - Info Model
==============================================

* Attendees:
	- ZackK
	- Ellen Stokes (IBM)
	- Doug Haigh (SAS)
	- Donal Fellows (UoM)
	- Sergio Andreozzi (INFN)
	- Steve McGough (Imperial)
	- Chris Kantarjiev (Oracle)
	- Chris Smith (Platform)
	- David Snelling (Fujitsu)
	- Laurence Field (CERN)
	- Andreas Savva (Fujitsu)
	- Hiro Kishimoto (Fujitsu)

	Minutes: David Snelling

* Summary of New Actions

	- AI-0607a: Doug or Chris S to look at the use cases template
	  and consider mapping the SAS use cases to this template. 
	  - Can be added to the wiki or mailed to the group.
	- AI-0607b: Doug to test if attributes on the Execution
	  Environment can address the need for boolean capabilities.

* Agenda Bashing

	- No Change

* House keeping:

	- Minutes approval from May 24: OK

* OGF/DMTF Work Register

	- See WR:
	  http://www.ogf.org/documents/liaison/OGF-DMTFWorkReg20061106.pdf
	- See Slides: https://forge.gridforum.org/sf/go/doc14581
	- WR presented to the GFSG Standards Council call.
	- They decided to discuss them with both the OGSA WG and the
	  whole of the GFSG.
	- Take a positive approach with respect to the WR.
	- Some tasks undertaken in Q1 2007, see slide 3
	- There is a partial challenge that we face as a result of
	  communication difficulties inside DMTF.
	- May also suffer from the Volunteer Army problem, like OGF.
	- DMTF have only weak processes for dealing with external
	  comments
	- E. Stokes needed to push inside DMTF to get comments into
	     the right place.
	- Schema submission to CIM are in process, but still requires
	  proactive external involvement.
	- The ballot process is very slow.
	- CIM releases three times a year, and we would like a late
	  summer data for all OGF contributions.
  	- E. Stokes has been promoting the submission in the DMTF and
  	     OGF.
	- Recommendations:
		- Improve the alliance partner process improved.
		- Proposal to delegate control of branches of the CIM
		  model to partners.
		- This would require some further developments
		  required in DMTF, e.g. access to validators.

	- Questions etc.
		- Hiro: How have you done this in the past, e.g. JCIM?
		  Some else covered this time period.
			It did become part of the model. There was
			more support from inside the DMTF for this
			work.
		- Donal: The approach seems like a good approach.
		- Ellen: The aim is to maintain DMTF's position as the
		  port of call for all models.
		- Ellen: The working groups are likely to push back,
		  but the higher levels of the organization will be
		  supportive.
		- Dave: Is this the set of slides that we are planning
		  to use? Yes.
		- Donal: A bit too much text, but are probably of for
		  a phone based call.
		- Dave: Use a DMTF slide template.

* Use cases: SAS Use Case:

	- See:
	  http://forge.ogf.org/sf/sfmain/do/downloadAttachment/projects.glue-wg/wiki/UseCases?id=atch4434
	- The use case focusing of the data model and not on the
	  submission/management side.
	- Aim to test the GLUE Schema.
	- Need for boolean capability resources. Needed for
	  matchmaking.
	- Need a status testing ability, but this may be provided by
	  BES requirements.
	- Need to find out what host a job is running on.
	- The GLUE schema currently does not plan to provide the
	  dynamic details of the final worker nodes.
	- This might be a disconnect in requirements, but the level of
	  detail needed seems low.
	- The question may just be the need to clarify what level of
	  the Grid is being exposed.
	- The schema itself is adequately populated and the use of it
	  may be sufficient to deal with these differences.
	- Can GLUE be used to expose this much detail? Yes,
	  roughly. Could be addressed to exposure profiles.
	- In the cluster use case there are plans to expose the
	  details of the work nodes (including dynamic information).
	- The interface for obtaining this information should be
	  available from the "head" node, even though the information
	  comes from the worker nodes.
	- The level of detail is a policy decision, and GLUE does not
	  restrict/mandate it use.
	- Recommendation that examples where only abstract summaries
	  are required.
	- GLUE does not, however, define this policy language.
	- Profiles, in addition to the schema itself, should be use to
	  refine required elements.
		- VO Profile has only high level information
		- Cluster Profile has lots of details about individual
		  nodes.
	- It may be possible to use the attributes in the Execution
	  Environment.

	- AI-0607a: Doug or Chris to look at the use cases template
	  and consider mapping the SAS use cases to this template. 
	  - Can be added to the wiki or mailed to the group.
	- AI-0607b: Doug to test if attributes on the Execution
	  Environment can address the need for boolean capabilities.

	- The schema is available for the wiki page.

	- Next Meeting in GLUE WG schedule. Please cross post to
	  OGSA-WG list.