OGSA-DMI phone conference minutes
=================================

18 July 2007

Participants
============
Mario Antonioloetti, EPCC
Allen Luniewski, IBM,
Michel Drescher, Fujitsu

Apologies: Steven Newhouse

Agenda
------
1) Early discussion (15 min)
    - Note taker assignment
    - Roll call
    - Telecon minutes approvals
        - 11 July: http://forge.gridforum.org/sf/go/doc14676?nav=1
    - Action item review
        artf5923 - "Undo strategy: 'Full undo'"
        artf5931 - "Split DONE state into two states: success and failure"
        artf5949 - "Consolidate and merge the two existing use case example
                   documents"
        artf5950 - "Capture security requirements and place them in the
                   security section"
    - Agenda bashing

2) The role of the DataEPR

3) Information contained in the Data Wrapper

4) Wrap up
    AOB

1. Early discussion
-------------------
- Michel takes notes
- Minutes approval
  -- 11 July: approved without changes
- Action item review:
  -- artf5923 reviewed and closed.
  -- artf5931 reviewed and closed.
  -- artf5949 reviewed and closed.
  -- artf5950 reviewed and closed.
  
2. Role of the DataEPR
----------------------
- The group discusses the role of the DataEPR and the data wrapper complex data structure.
- The group conjectures that defining the complex data wrapper is necessary.
- The group postpones the decision whether to envelope that data wrapper in a wsa:EndpointReference 
  or not to a later point in time when more information is available

3) Information contained in the Data Wrapper
--------------------------------------------
- It contains a sequence of dmi:Data elements
- dmi:Data element has two attributes 
  -- protocol URI (defining the protocol to use for the data transfer itself)
  -- protocol dependent URI (to use for passing into the invocation of the data transfer protocol).
     --- this URI is a translation/mapping of whatever eveloping abstract data structure (e.g. EPR)
	     defines the file in question for movement.
- Michel proposes an additional attribute, called for now "style", that describes the data storage's
  capability for that particular data transfer protocol, whenther it can operate in "push", "pull", 
  or "3rd pary" mode.
  -- For example, a source denotes a "push" capability for FTP, and a sink denotes "push" for its own 
     FTP capability as well, then DMI must rule out FTP as a possible data transfer protocol between 
	 those two storage locations.
  -- What about the value 3rd party?
     --- This denotes a setup where data channels are opened between source and sink, and two control channels
	     from a controlling agent to source and sink, respectively.
- The group concludes consensus on the layout of hte complex data wrapper element, except for Steve's 
  confirmation on the proposed new "style" attribute.

4) Any other business
---------------------
- Michel has to step down as WG chair, and even possibly as an active member of the group.
- The group discusses several possible consequences to that decision.
- The group agreed that Michel should contact Ravi Madduri, and the Data ADs, and then put it on the 
  mailing list.