OGSA DMI Skype Telecon - 22 April 2010, 8am BST/GMT+1 ----------------------------------------------------- Present: Mario Antonioletti, EPCC [MAA] Steve Crouch, University of Southampton [STC] David Meredith, STFC [DM] Agenda: 1. Agenda bashing 2. Actions review 3. Progress update 4. Opening up an interaction channel with EGI for data movement -- Response from Steven Newhouse begin_of_the_skype_highlighting end_of_the_skype_highlighting about OGSA-DMI work (esp. bulk data copying) 5. DMI strawman -- Feedback required from group -- Define timeline for merging DMI strawman into func.spec. 6. Phil.Trans.A DTS paper (deadline for camera-ready version Monday 26th) 7. Preliminary OGF29 discussion 8. Planning 9. AOB 10. Date of next meeting New Actions: ------------ [STC] Set up a Google poll for next call - and post to the OGSA-DMI list Completed Actions ----------------- [DM] Clarify a few points on strawman slides (inc. JSDL approaches, in-line vs element references) [STC] After strawman slides updated, send slides to S.Newhouse for feedback and canvassing requirements [SM] To send round link to DMI implementation on SourceForge Ongoing Actions: ---------------- [STC] Draft an extension to OGSA-DMI charter to encompass bulk data copying activity - Edit the version on the GridForge website - the master version [DM/SM] Look at functional spec and evaluate necessary changes w.r.t. strawman - And post up an example [STC] Query Ravi about involvement with OGSA-DMI to engage Globus - and get a technical contact [DM/STC] Ask Andrew Grimshaw and Mark Morgan begin_of_the_skype_highlighting end_of_the_skype_highlighting University of Virginia about: - Element references with BulkDataCopy - DEPRs and EPRs - opinions on how best to proceed - Send presentation to them but summarise in bullet points in the email; include corrorobative statements from S.Newhouse [DM/SM] Improve the strawman document in collaboration - Catch up required to merge changes and discuss points - Need to change functional spec - Couple of days effort - Feedback is important [DM] Upload strawman document to the gridforge in order to have more visibility to other participants - Effectively the presentation uploaded to OGF28 is a first draft - Functional spec - change directly? 3. Progress Update Consensus to table the progress update until next time. 4. Opening up an interaction channel with EGI for data movement S.Newhouse email reply circulated to list with his permission... DMI - abstractions include sources and sinks; which could be anything including databases. Opportunity for scope creep - would we want to consider transformations; from one data type to another? iRODS - how does it fits with DMI and DTS? M.Atkinson - moving computation _to_ the data. What are data sources and data sinks? Perhaps we should look further into classifying these? e.g. OGSA-DAI, SRM, etc. at the moment, goes back to EPR/DEPRs and how to specify the metadata within other data within it? e.g. for DBMS - perhaps could put in SQL direct to access data? Or, could have an abstraction mechanism to group data together at the source end and use that within the DEPR/EPR. Discussion of how to access multiple files within a source; specify DBMS data access requirements in the transfer requirements. Keep tabs on the EMI/UMD process for adopting software for the DTS for its possible inclusion (virtual or otherwise) within the UMD. Data requirements from EGI - data session at the EGI tech forum; perhaps we should have representative at the forum from OGSA-DMI. 5. DMI strawman Consensus to table the DMI strawman discussion until next time. 6. Phil.Trans.A DTS paper DTS paper - D.Meredith to trim down the paper and send out a revised version hopefully today for review. 7. Preliminary OGF29 discussion S.Crouch very likely going so far. Need to do some preliminary work to bring others in. Ask A.Grimshaw about his attendance (when emailing about EPRS/DEPRS and element references). 8. Planning 9. AOB 10. Date of next meeting New action: [STC] Set up a Google poll for next call and post to the OGSA-DMI list.