Minutes of the Data Design team telcon, 2nd November 2004 0) Summary of actions - Dave to ask UVa folk for response to GFS. - Metadata scenarios due by Nov 16th. - Strawman architecture due by Nov 30th. - Dave to mail architecture notes to a wider group. 1) Early discussion (5 min) Roll call: Manuel Pereira, IBM Ted Anderson, IBM Leo Luan, IBM Mario Antonioletti, EPCC Simon Laws, IBM Shannon Hastings, OSU Susan Malaika, IBM Dave Berry (NeSC, note taker) The minutes of the Oct 27th telcon were approved. We added an agenda item on the strawman architecture and a restatement of our overall goals. 2) Update on outstanding actions (5 min) - Manuel to post new version of GFS VFDS spec by end October (if possible). [This has not proved to be possible as some changes still require review. Manuel would like to discuss the proposed changes on the call] - UVa team to read the GFS specs. [Status unknown] - People to register for gfs-wg mailing list for notifications - GFS team to review UVa documents. Manuel to be point of contact. [Done] - Dave to send reminder and pointer to naming requirements documents [Done] - Other assignees to produce use-case scenarios for use of metadata. [Ongoing] - Andrew, Allen and Dave to continue with a strawman architecture. [Progress has been glacial. It would probably help to open out the discussion] - Dave to send details of December OGSA F2F meeting [Done] - Dave to schedule discussion of data movement for a later telcon [Ongoing] 3) BasicFiles and Contexts - GFS: require clarification on how the Contexts scheme map onto the three-level naming scheme. Are Contexts mapping human-readable names to abstract names, or abstract names to physical names? If they are mapping human-readable names to abstract names, where are the abstract names resolved? - VFDS document clearly deals with a 3-layer naming scheme. - One of the focusses of VFDS is to address the traversal by humans of the hierarchic namespace. Need conventions for the arrangement of the hierarchy; VFDS provides this. - VFDS facilitates performance in dealing with both location independence and 3-level naming resolution. - One of the substantial revisions to the VSDF document is to incorporate WSRF. This will be done at two levels. First in the specification of use (i.e. use WSRF as the communication mechanism. Second in allowing WS-A/WSRF endpoints in VSDF junctions. - We need the UVa folk to be present for more detailed discussion. Dave will contact them to encourage a response and ideally a meeting. If the overlap between OGSA and VFDS is close then it may make sense to for the relevant OGSA people to join the GFS meetings. 4) Update on Metadata scenarios Simon has received some inputs; others are outstanding. Done: Some of - Access, Discovery, Composition Awaiting: Replication/Caching, Federation, Ontologies, Delivery We agreed a deadline of Nov 16th. 5) Strawman architecture Dave gave a recap of the aims of the design team: - to outline the services required to implement the data capabilities of OGSA - to show how these services collaborate to implement a range of scenarios - to engage and stimulate the relevant working groups Malcolm Atkinson and Kostas Karasavvas have produced some work on data connection, as part of the metadata work and as part of the EU NextGrid project, which will feed into the architecture design. Other scenarios from the metadata work should also feed into the architecture design. Dave and Allen have has a very short mail exchange about whether structured data access (i.e. of the sort specified by DAIS) fits naturally into a file-oriented architecture such as EGEE, or whether structured access is a higher-level notion. We need a straw man by Nov 30th, so that we have something to take to the OGSA F2F. The straw man doesn't have to be correct; it will be the basis for discussion and improvement. 6) Wrap up DONM: Nov 16th for metadata discussion. Possibly earlier for UVa/GFS discussion?