OGSA-WG Teleconference -- 16 October 2003 ========================================= * Participants Bill Horn (IBM) Andrew Grimshaw (UoV) Hiro Kishimoto (Fujitsu) Fred Maciel (Hitachi) Takuya Mori (NEC) Andreas Savva (Fujitsu) David Snelling (Fujitsu) Ravi Subramaniam (Intel) Daniel Templeton (SUN, DRMAA) Jem Treadwell (HP) Apologies: Dave Berry (UK NeSC) Minutes: Andreas Savva * GGF9 ** Minutes approval - GGF9 minutes (pdfs) uploaded to gridforge can't be read due to included asian fonts (again...) - Acrobat asian font pack can be used to work around this problem. Works on Linux too. Ref: http://www.adobe.com/products/acrobat/acrrasianfontpack.html - Action: Redo pdf conversion >Andreas - Agreed: Postpone minutes approval since people couldn't read them. - For the last session (platform) only have Fred's minutes. - Action: Ravi to send his minutes for the last session (platform) to Andreas ** Overall impression of OGSA sessions - Overall agreement that the sessions were successful. They were informational not just for the OGSA-WG but also for the other participating groups. (Making them aware of each other's work.) Some overlap became obvious; it should help our factorization work. - This was our first try so by necessity we couldn't go too deep. It was a good starting point but we should focus on more detailed agenda next time. - Agreed: It is important to keep the conversation going with the other groups. - Comment: If we really manage this process well the branding comes automatically. ** Branding - OGSA branding was discussed in GFSC. GFSC was sympathetic, but there was concern about OGSA-WG taking on too much {responsibility, namespace}. - (People can say they are using OGSA freely; It is more difficult to talk about compliance.) - Agreed: Be vigilant about people using the term 'OGSA' and talk to them if we think they do not use the term correctly. - Agreed: Write up how we expect people to relate to OGSA (and push to GFSC too). - Action: Ravi to do first draft of this document. Comment: Ravi seemed to have a more technical perspective (define plugpoints) but one example classification that came up was: 1. part of ogsa 2. compliance with ogsa 3. building on ogsi (but not necessarily ogsa): ogsi-compliant Agreed that we can't just let people say "OGSA like" unless we can define what "OGSA like" means in concrete terms. ** Session summary Worked through half of the one line summary sent out before the teleconf by Andreas and added comments. Postponed the latter part since some people had to drop out early. *** Data session **** DAIS - Alignment with OGSA doc Globus+IBM and DAIS worked things out prior to GGF. There was no obvious difference between what's in the OGSA document and the DAIS presentation. Comment: DAIS started from RDB but has grown to cover various data sources, e.g., XML and files, hence the overlap with GFS-WG --- see further down. Agreed: Keep watching. **** OREP - Building on top of what DAIS was; a bit out of sync at the moment. Might take a bit to come round. Because of their relation with DAIS we expect that they'll move over to current DAIS abstraction. (No action on our part.) Agreed: Keep watching. **** PA - Not clear yet if OGSA data abstractions are what they need. (Waiting for more refinement.) - They don't claim to be OGSA at the moment. - They are well advanced; they have stuff working. It will take them a while to 'come round'. - Agreed: Let them finish and then look into possibility for an ogsi/ogsa re-factoring. In any case it doesn't affect us too deeply --- mainly implementation from our point of view. **** GFS - Not clear that they want to build on top of DAIS. (SOA too complex for them at the moment?) - They seem to have narrower focus than desired (by us). Not well integrated with what's happening. (They want to support POSIX I/O APIs and (?) global name space services.) - Their 2nd BOF voted to make it a RG (23 for, 8 against) but that doesn't seem right. There are already products out in this space so it is not a new topic. - Agreed: Look at them again when/if they become a WG. *** PE session **** GRAAP - Building on top of OGSI. Alignment with OGSA doc. - Agreed: We have to watch this space. In particular we need to really understand what agreement is, as it is creeping through the entire architecture. We should begin by highlighting all concepts of agreement that *we* are interested in and making sure abstractions are strong enough to be used throughout. - Comments: - It looks like there are still many issues not addressed. Work is in initial stages; needs refinement. - DAIS using agreement. Not sure if they are also defining agreement terms for data.(?) **** JSDL - Presentation not just on job submission description. Willing to explore OGSA role but also looking at other grid environments. - Presentation gave the impression that they are trying to tackle more than what their charter says. They are not. It is just job submission description. - GRAAP is strongly involved in the WG. DRMAA is also interested in it. - JSDL would also like to be used in non-ogsa/agreement settings. - Agreed: Keep watching. ** DRMAA - Interested in exploring role in OGSA. - There is no overlap between DRMAA & JSDL. Think of DRMAA as having defined a job template that can use jsdl inside. - So far the focus has been on legacy stuff. Finishing with C, java binding and may focus next on gwsdl (ogsi) binding. - Agreed: They are not affecting the architecture but the implementation. - Agreed: Useful to continue a dialogue. - Action: Andrew to look into bringing them into PE/Provisioning subgroup. [Postponed discussion of remaining sessions for next call.] * Plan and schedule ** Telecon time slot in November and after (Day-light saving time will end soon.) Resolved: Use the same timeslot [4:00pm CST], which means that Japan teleconf time becomes 7am and no change for other people. ** F2F meeting date/place Proposal: Do 2 F2Fs before GGF10. One in December and one in February. Proposal: Do both F2Fs in California. Andrew suggested SDSC as a possible venue, at least for one of them. Action: Hiro to send out spreadsheet and get people's preference for Dec&Feb. Comments: - Fred prefers week of Dec.8 (will already be in US) - Week of Dec.8th is exam period in US. Andrew can't travel. - Bill, Andrew prefer week of Dec.15. - Tail-end of week of Dec.8 may be a possibility. ** Documents review schedule *** Missing texts Action: Review trackers and submit your text >ALL. And in particular Andrew for the Glossary section. *** Jeff's comments Started going through these during GGF9 during a brief non-public session. - Agreed: Get Jeff on the call and continue working through them. *** Other comments (Jem) - Jem posted some questions; mostly answered by Ravi. - Proposal: For wordsmithing stuff use Word's tracking feature to update and let next person with the pen accept/reject. (OGSI approach) Comment: Accepting changes leaves comments in so deeper issues can be carried in the document. - Action: Jem to finish up and send document to Hiro (or upload). ** Sub-group status **** Program execution subgroup (Andrew) - Action: Andrew to send out minutes from GGF9 subgroup private session. - Proposal: Change subgroup name from PE to something else. - Andrew can't make next couple of PE phone calls as he will be travelling but can make the 4 calls after that. - Agreed: Cancel the next 2 PE teleconfs on the understanding that Andrew commits to driving the process forward in time for the Dec. F2F. - Action: Andrew to draw up series of PE agendas(4) leading up to the Dec. F2F. **** Logging system subgroup (Jim) - Teleconfs gave good feedback; ready to update document. - XML seems to be holding up. - Next step: Position logging wrt to ogsa fabric services (messaging, events, etc). - What to do with events? DAMED (defunct) used to do grid events. There are (at least) 2 specs in this area at the moment: - WS-Events (HP) - Common Base Events (IBM) - Rejected: Form event subgroup. We already have 3 teleconfs a week and it is getting difficult to get enough people on the calls. - Agreed: Keep the events discussion within the logging subgroup. - Action: Bill to send out annoucement and get more people involved in logging subgroup. Comment: Andrew would like to join the discussion on events but he won't be available next week.