DAIS Telcon 22/04/04 ==================== Chair: Norman Paton/Dave Pearson Minutes: Mario Antonioletti Present: Mario Antonioletti, EPCC Dave Pearson, Oracle Simon Laws, IBM Norman Paton, University of Manchester Vijay Dialany, Southampton University Susan Malaika, IBM Agenda: Actions from the previous telcon Composite Requests (Dave Pearson/Greg Riccardi) AoB New Actions: ------------ [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. Actions Resulting from previous Meetings: ----------------------------------------- [Simon] Ensure that DAIS liaises with other GGF WGs in a systematic fashion. (Identify persons that are interacting with other GGF groups and document this). Dead Actions ------------ [Dave] To type up names of folks that said were interested in files at GGF9 and pass these on to Neil. Done shortly after the meeting. ---- Outstanding actions were reviewed. Dave produced some slides demonstrating patterns about composite requests. Slides can be found at: http://forge.gridforum.org/projects/dais-wg/document/DAIS-WG-Topic-Composite-Requests/en/1. ... Presentation followed the slides fairly closely. Mario: do composite requests affect one service or multiple services? Dave: Assumed that you could compose a service that would handle this kind of request ... Norman: an argument for not encompassing multiple services is that there are already mechanisms for doing this - e.g. workflow - if composite requests are to be in the DAIS scope it should not have any aspect of orchestration. Dave: I was thinking in terms of a single service. Norman: have a diagram which has inputs, outputs and transformations go into a data service ... you have to add value over that provided by the mechanisms to group sequence of requests already supported by a data resource, e.g. SQL 99 ... the next step up, in a service context, is that you might want to do the things that Dave describes, and be able to pull in data, perform transformations and ship data out. That is the kind of thing that you would like to do. ... Mario: is what you describe not in the infod territory? Norman: No, infod is publish subscribe while this is still in the request-response model. Characterise request-response as being in the DAIS territory. Susan: We should publish a precise definition as the infod working group might have a different view. Dave: ... we would see infod as being higher-level than what we are trying to do. ... Susan: everyone might not agree with this difference. Vijay: Dave do you consider staging of requests? Dave: have assumed that you submit a batch of requests then ... Vijay: what I meant is that if you place a request on the service then someone else can fire off that request and get the data back ... something that you might want to do as an administrator... Dave: I see this as one of the advantages of composite requests. Vijay: another thing about transactional context being all or nothing .... Dave: certainly for the administrative domain do not need acid transactions but for bulk loads you might want to have a transactional context... Vijay: can we make a statement that all the local requests can be involved in a transactional context ... Dave: you have to be clear of the conditions Vijay: for local transactions we should assume transactions that the local data resources may or may not support the given types of transaction ... Dave: yep that would be a good approach ... once you get into a batched requests then you have to ... Vijay: ... we have to be clear that this only allows localised transactions. Dave: I think that is a very important point Vijay. ... Discussion about what constitutes a logical unit of work and Dave proceeded to go through the slides. Discussed patterns - Jim Gray's work with the astronomy examples. In writing these examples out it has become apparent that there are different ways of defining the levels of units of work and how you define these might be application driven. Dave concludes his presentation. Norman: Dave has provided a characterisation of different patterns for levels of work ... there are two proposals on the DAIS table dealing with composite requests: Greg's and the one provided by ogsa-dai. We would like to get interested folks to come up with a community practice document for some future GGF. These slides could be used to validate such a piece of work ... maybe what we do with this just now is note the proposal and when people come forward to take work on composite request forward they could use these slides.... Dave are you interested in working on such a proposal or is this what you would like to see included in any further work on composite requests? Dave: this could be fed into the use cases that Dave Berry has been putting together for their ogsa data service (architecture) work. Norman: what do we do next? ... the people that are holding the token here are Greg and Malcolm ... need to get them to come on a telcon and get them to propose how this can be moved forward. Is there anything else that we can do? Dave: all we can do is tell the folks at GGF what has been going on in this area ... and try to encourage participation ... there is nothing here that could not be done at the orchestration level. It's a question about the granularity of service that people want to define. Norman: this is another set of useful scenarios for composite requests Dave takes an action to email Greg and Malcolm to try to find a slot as to when they might be able to join a telcon and try to get a proposal on how to progress otherwise we will not be able to include this in our future GGF charter. At the moment there is a demand but not enough supply. AOB --- Norman: DMTF - Susan - attempts to set up a telcon ... Larry has been silent ... Susan: it appears that it's going to be difficult to get both of us on a call ... maybe we might be able to just have one of us? ... need to try make sure that Tod is on the call (chair from oracle on the data area working group). ... I'll bug them. Norman: it will not bother me if I can't turn up to a call. Norman: Simon - there was discussion about batting out the slides about the mappings ... any progress ... Mario comments that the slides presented at GGF9 have been up on Grid forge for a while (shortly after ggf9) and that he recollects that it was the document they were working on and not the slides. Simon: we hope to distribute a copy of the document by the end of next week ... Norman: has there been any progress on the specs? Simon: have written half of the words on the core spec, Mario is massaging the words ... have been doing the WSDL at the moment. The XML spec is awaiting the WSDL and the core specs. We can get Amy to catch up on this once this is done ... have almost finished the WSDL. Norman: I may be able to look at some stuff over the weekend of the first of May. Simon: our goal was to have a draft of the docs ready by the end of next week. Mario: ... my contribution to the core spec at the moment has tended to produce more question than contributed text ... ... Simon: I'll bat the WSDL to you mario soon to comment... Mario: what about the relational spec? Simon: that is Susan/Brian ... presently producing an example based on the relational spec ... have not produced any words yet. Brian is worrying about various bits and pieces. ... Norman: I'm comforted ... Susan: going to be attending the OASIS WSRF meetings next week - let me know if you have any questions ... OASIS have all day meetings .. a whole day on wsrf and a whole day on ws-notification ... most of the OASIS activities run in parallel. Mario: is the WSDL WSRF based? Simon: yes, I'm doing the WSRF mapping. Vijay: do you know when the infod people meet? Susan: there is a telcon tomorrow ... we are reviewing the charter and a couple other things ... it was posted to the dais and the infod mailing list .. did you not get it? Vijay: do not recall? various other people commented that they had received the message about the infod telcon (Simon forwards the infod telcon announcement to Vijay). Susan: ogsa - dave berry's group - has an extended call next friday and are going to try and have an e-meeting and they are going to work through scenarios and use an electronic white board ... want to prepare the infod people to contribute to that effort. Mario comments that Greg had posted an emails asking about the telcon number but as he was taking minutes he did not see it. Dave will post a reminder about the telcon numbers. Susan comments that Larry Flon had posted a message about a DMTF/CMM telcon.