Minutes of the OGSA Data WG telcon, 15th March 2006 0) Actions arising * New actions - Allen to book GGF17 session. * Ongoing actions - Dave to forward old message about interface description. - Stephen to add a paragraph to the architecture document about reasons for naming, if still needed after Allen's revisions. - Stephen to add interface names to the scenario diagrams. - Ann to check the Access section for suitability for file replication. - Neil to update the ByteIO subsection. - Section authors review security aspects of their sections. - Everyone to talk with your contacts to get more participation in this WG. - Allen/Dave to contact David Martin & Hiro Kishimoto to get reviewers & expert participation. 1) Early discussion Note taker assignment * Roll call Dave Berry, NeSC (note taker) Allen Luniewski, IBM Mario Antonioletti, EPCC Stephen Davey, NeSC Chris Jordan, SDSC The minutes were approved. 2) Action report - Dave to forward old message about interface description. [Ongoing] - Dave to mail Reagan re SRB. [Done] - Dave to raise the question of URIs for data formats (etc) at the GGF16 data area meeting. [Done, but matter is unresolved] - Stephen to add a paragraph to the architecture document about reasons for naming, if still needed after Allen's revisions. [Ongoing] - Stephen to add interface names to the scenario diagrams. [Ongoing] - Ann to check the Access section for suitability for file replication. [Ongoing] - Neil to update the ByteIO subsection. [Ongoing] - Section authors review security aspects of their sections. - Everyone to talk with your contacts to get more participation in this WG. - Allen/Dave to contact David Martin & Hiro Kishimoto to get reviewers & expert participation. 3) GGF16 Report * Our session 1: presentation of status & responses. In this session we discussed the architecture document and the scenarios document. - Data transfer: Reagan noted that we may need to synchronise metadata transfers with the corresponding data transfers, particularly when the metadata is held in separate storage. - Data storage: Jens noted that the terms used by SRM (durable, volatile, etc.) are confusing because they imply a policy about archiving the actual data. This is being addressed in the GSM WG. - Data federation: In response to a question, we confirmed that We do support federation of federations. - Metadata catalogues: There was a discussion about how metadata catalogs are linked to file stores. In our architecture, this role is filled by abstract names. - Dieter asks why we use names as intermediaries in discovery mechanisms, rather than just passing the characteristics of the required data to the access mechanisms etc. We think this is too ambitious for sa standarisation group. Dieter said that InfoD are discussing this. - We discussed how to create a hierarchic namespace in the replication scenario. The issue is which names are stored in the replica catalogue service: human names or abstract names. - We clarified that the replication scenario is intended to handle databases as well as files. Also that this is not the same as storing timestamped copies of data or copies of versions (which perhaps should be separate scenarios). - In the pipeline scenario, we need to clarify that the rendering service also controls the data transfer. - We clarified that in the federation scenario, the data integration service performs format translation and integration. - When discussing the storage scenario, we were asked about storage scavenging. We thought that would be abstracted behind the SRM interface. - Reagan offered to provide some extra scenarios. * Our session 2: Grid File Systems, provenance. - We discussed the Grid File System as a possible scenario building on the services of our architecture, understanding what was needed from each side. We undertook to write a draft scenario encapsulating our understanding. - We also discussed a provenance architecture. This is essential using data services for a particular end. We will produce a scenario to address this. * Other sessions of note - The data movement BOF showed enthusiasm for the proposed working group and seemed to make good progress. This is good news. - The OGSA F2F meeting did not discuss data per se. It did make some progress on the Information Model. Marvin Thiemar presented a proposal to write an HPC OGSA profile, which met with approval. This is similar in spirit to our scenarios, although not restricted to one area (e.g. data, execution,...). The mechanism is likely to be a profile akin to the WS-I profile or OGSA WSRF Basic Profile. 4) Progress & planning Allen to book a GGF17 session as a status update. At the moment it isn't clear how to structure this. Dave & Stephen are not planning to attend GGF17. We have been invited to attend the SRM telcon on the 29th March. We agreed to attend this in lieu of our own scheduled call for that week. We discussed topics for future meetings, using Stephen's list: 1. Review of latest data arch doc. 2. Interfaces for arch doc. 3. Interfaces for scenarios doc. 4. Relationship with SRM (GSM-WG) 5. New GFS scenario. 6. New Provenance scenario. 7. GGF17 planning 8. Discussion of DAIS (and ByteIO?) specs - and comments 9. Metadata catalog scenario & services (cf. SRB) 10. Review of (final?) arch doc. 11. Review of (final?) scenarios doc. 5) Wrap up DONM - with SRM on March 29th.