DAIS Telcon 22/04/04 ==================== Chair: Dave Pearson Minutes: Mario Antonioletti Present: Mario Antonioletti, EPCC Vijay Dialany, Southampton University Simon Laws, IBM Thomas Soddemann, RZG Norman Paton, University of Manchester Dave Pearson, Oracle Dave Berry, NeSC Amy Krause, EPCC Brian Collins, IBM Allen Luniewski, IBM Agenda: Actions from the previous telcon OGSA Data Services Document (Allen Luniewski) Object Database Access Services (Thomas Soddemann) OGSA Data Architecture as it affects DAIS (Dave Berry) AoB New Actions: ------------ Actions Resulting from previous Meetings: ----------------------------------------- [Dave] Email Greg and Malcolm to determine a time when they might be able to join a telcon in order to see how to take the composite request framework forward. Have sent an email - have not had a reply yet. [Simon] Document the DAIS folks that are liaising with other GGF WGs. Document who is. ----- o OGSA Data Services Document (Allen Luniewski) [ Dave = Dave Pearson, Dave B. = Dave Berry] Dave gets Allen to lead the first part of the call. Allen: thanks Norman for getting the Data Service work moving again. Try to get issues that are outstanding on the document. Gathered issues from the last GGF that were noted. - The document was changed to use WSRF. - A question about end point references. The intent is to EPRs to direct clients to the appropriate resources and this is not particularly well described. - A question about placement at the dropped services level - to get data and services at the right place. - A naming issue. - ? - A workflow - 3rd party delivery and how does it relate to infod and workflow. - Comments inlined in the document from Norman Paton and someone at IBM called Mike Swanson. Dave: has anyone got any more issues? Mario mentions that grid forge might be used to track issues with the ogsa data service document publicly although this has not been totally successful at tracking issues with DAIS documents in the past. Mario: has the doc changed since GGF10? No. Dave: wants to clarify the role of the management portType which is used to manage the virtualization but what is used to manage the underlying resource. There has not been any further development on that although it has been discussed in the past. Simon: did the data description go in the last version of the document? Norman: the previous version of the document had data description but in WSRF you do not require to have a separate portType to do data description ... Allen: I did not see the need for an explicit portType for description. Simon: should really consider these as categories and not portTypes .. the same thing applies to Factories. Allen: that is right. Simon: we might some things about data service factory but that is not necessarily an actual factory Discussion about whether a factory is needed across portTypes but no one has made a case for this yet. Simon: issue relating to delivery, events and notifications which started the infod effort off ... is there any need to say anything specific about this in the data service document? ... we don't know if an extra layer need to be added on top of this.... Dave: you think there should be a statement about this? Simon: we should think about this ... we had a previous discussion on triggers. Norman: something relating to that ... aspects of the document could cause it to be interpreted as a pattern for all data services in a Grid a framework. If you then ask if it accommodates all types of data service you find that it does not. Take the infod group who were originally asked to go and determine where they belonged in this framework and they found that they did not ... it's not that this framework is not useful but rather that it does not encompass all types of data service. So, if we are in a context where the data service document is seen as providing the functionality and DAIS as providing the top level portTypes then that's fine but there are other data orientated services that others might wish to come up with that do not fit in this framework, e.g. data movement, data replication, etc and that should be discussed in the document ... so data orientated services might not fit in ... this is not made explicit in the document yet ... it might be worthwhile flagging this. Allen: approach is that is good for certain things but not everything. Norman: this should be made more explicit ... about WSRF ... you should stick with the WSRF orientation ... as a group we are focusing on WSRF and discussing other frameworks in the work being undertaken by Simon, Savas and Sastry - the infrastructure discussion will take place under this banner and stop the discussion being replicated at the various other sub groups .. I invite people to comment on this position. Allen: I agree with that .... the Data Service document is not very reliant on using either OGSI or WSRF .... if these were to disappear most of the document would remain the same. Dave: Allen is there anything else that you would like to raise? Allen: will put the issues up on Grid Forge .... but then how do we resolve them. Not sure if we should use this call for this. Norman: but you should bring issues to this meeting ... reason for bringing Allen into the call is to try to discuss what happens to the OGSA Data Service document ... we need to think about mechanisms for resolving these issues ... we can use Grid Forge, this telcon and the mailing list ... could also use a f2f and GGF to move things forward. ... feel free to use all of these mechanisms. ... Could also organise telcons in particular to discuss this document. Allen: once I get these up on Grid Forge will send something out to the mailing list to try to begin the discussion. o Object Database Access Services (Thomas Soddemann) Thomas: had a telcon last week. There were only a few people there but nevertheless had an interesting discussion going . Norman put the minutes out to the list. Went through groups that need OO database access .... the intelligence community came up which I had not thought of before. Objectivity has customers in the intelligence community with databases in the 12Pb region ... really big databases. Covered other work of relevance to this effort. Covered some projects from where we could get ideas. Discussed query languages - the most appropriate query language might be JDO although this is possibly a bit Java specific but nevertheless was judged to be the best thing around that we know of. Have started to write some documents with some use cases which will discuss why OO database access might be useful. We will have another telcon in about 3 weeks which Norman will organise. Norman: might be worthwhile to add that Leon could contribute several days a month and Thomas can add several days a week. Will try to get a draft recommendation for GGF12 and a draft recommendation for GGF13. Will try to see if any other OO database vendors might be interested to contribute so far we've had quite a positive outcome. Dave: CERN use objectivity databases ... Thomas: I heard that they were moving to Oracle 10g Dave: yes, but they'll still have large amounts data which will not migrate to Oracle 10g for a while in Objectivity databases. Thomas: the Wendelstein 7X Fusion experiment will be using objectivity and they expect about 1Tb of data/day to carry through .. Dave: Peter Kuntz might be able to put you in touch with other folks at CERN that may have interest in this. Norman: it's inching forward ... would be good to get more people involved. Will have a heads up at the next GGF. Will not try to put anything explicitly in the charter until after the heads up but it's going forward well thus far. o OGSA Data Architecture as it affects DAIS (Dave Berry) Dave B.: background - trying to get a much better list of services and a better data architecture for OGSA. What is there at the moment is taken from the OGSA Data Service document ... but there are a whole bunch of services that are needed that are not described in any detail. Up to this point have been collecting use cases and getting people involved. Having a 3 hour telcon scheduled for tomorrow to talk about the use cases and try to come up with an architecture. need people from DAIS to be involved - which they are - and others to review what comes out from this effort. There has been some discussion between Allen and me to clarify the remit between the design team and the data service document people. We want to make sure that changes that happen in the DAIS OGSA Data service document are taken into account into the OGSA document. Peter Kuntz and Ian Baird have been recruited. Andrew Grimshaw is also involved. Dave: Any questions? I will be contributing some use cases for the commercial case .... the output from this effort will be useful for validating the work that DAIS is doing. Dave B: keen to know if there are any constraints that arise from DAIS that are applicable to the OGSA work. Dave: I would like to see composite requests being dealt with - the difference between a pipeline and workflow - any work that can be done in that area would be of interest. Dave B.: I have a personal interest in this. Keep on nagging Malcolm Atkinson to contribute ...but he's busy. Intension is to submit the OGSA document at GGF11 but the data part may not be in a good state by that time. There is a f2f at ANL the 11 & 12th of May ... if anyone is interested then they should try to contribute to make sure that data is an important part of the OGSA document... if the document gets put forward at GGF11 then need to discuss whether data services are correct. Dave: what are the time scales? Dave B: OGSA is supposed to be the specification for the Open Grid Service Architecture ... not just DAIS .... would like input from other WGs. Vijay: wish to understand the remit of the OGSA WG as they have adopted WSRF ... OGSA had a design philosophy and then they went to WSRF ... should data requirements be in line with the WSRF spec or in line with the old OGSA specification. Dave B.: the new doc is being re-written to take into account WSRF. There may be a revision at some later point. Viajy: one of the problems I see is that they have adopted a new spec - I'm not sure what the process that was used... Dave B.: it was discussed in the OGSA mailing list and telcons ... that's how the decision was undertaken .... Vijay: raises concerns if that group is not stable Dave B: not intending to hop around ... Vijay: just trying to understand the process. Mario: the fact that the OGSA document is going to be submitted for GGF11 means that the OGSA doc will be based on WSRF which by that stage will not a standard? Norman: the OGSA document will not be a standard ... will it be a recommendation? It can't be a recommendation if WSRF is not a standard ... ? Dave B. : not sure ... Dave: WSRF will not mature for at least a year ... Dave B.: I will try and find out what the process is going to be. Dave: if the doc is going to be submitted at GGF11 and if it's not going to be a standard then there will be a chance to change the data bit at some future date. Dave B.: will try to find out what the status is going to be and what versioning is going to be used. Brian: are there are any plans at GGF11 during the OGSA session to focus on the data parts of that document? Dave B.: yes there are ... Brian: that's good news ... there may be conflicting sessions from other relevant WG meetings if the DAIS people want to go to that ... would be nice to ensure that it does not conflict with any data focused areas. Norman: I have asked that DAIS sessions do not conflict with OGSA sessions... OGSA and DAIS get scheduled first so we should be in with a good chance.... Dave B.: OGSA Data session is being discussed for 30(?) mins on Tues at 10:30 ... there are also some design team session where data gets discussed for 45 mins. o AoB Brian: Neil Chue Hong and I have been working on a file group charter BOF for GGF11 have had some comments back from Peter Kuntz and Martin Westhead which then allowed us to produce a second version ... the second version has gone out to the list for comment. Peter Clarke has kindly offered to host a telcon to see if there are any overlaps with other groups ... final submission is on the 14 or 15 May ... would encourage that anyone has an interest in this to please comment. Dave: reminds people of the planned content of future DAIS meetings.