DAIS Telcon 07/09/04 ==================== Chair : Norman Paton Minutes: Mario Antonioletti Present: ------- Mario Antonioletti, EPCC Brian Collins, IBM Simon Laws, IBM Susan Malaika, IBM Norman Paton, University of Manchester Savas Parastatidis, University of Newcastle Allen Luniewski, IBM Agenda: ------ 0. Minutes/Actions. 1. Discuss sessions at GGF12. 2. Discuss object database. Actions: -------- New: --- [Norman] Get in touch with Dave Berry to see whether any input is required from the OGSA Data Design Team Effort. [Mario] To post a message to the DAIS mailing list that the GGF12 version of the DAIS specs are up on Grid Forge. [Simon/Brian/Amy] Prepare a slide set for GGF12 to discuss any outstanding issues for the DAIS specs and ensure that Grid Forge is up to date. [Simon/Savas] Prepare slides for the DAIS mapping session. [Simon] Contact the OGSA Naming Group regarding the DAIS mapping session. [Brian] Post the files charter to the DAIS list and invite comments. Outstanding ----------- [Dave] Distribute the executive summary document together with Mario's comments and try to attribute where the document excerpts came from. [Norman] Arrange a get together at GGF12 for those interested in an object data service realisation. --- Norman: Have three DAIS sessions planned for GGF12: 1. Discuss the specs, Dave will chair. An overview of the specs - Norman Discussion of issues. Need to discuss the format regarding the discussion of issues. Brian and Simon have posted the resolution of issues. Simon: should have all issues in Grid Forge. We need to go through these. There is not a huge number but there are some and there are some interesting ones. Can be difficult to illicit discussion though. Norman: might be good to try and work up to the issues and include some slides. Simon: ... have disconnected operations ... data management and what that means ... one about metadata - a pending one that we are not sure belongs in DAIS and there are a few others that we could talk about. The hot ones are the basic access and disconnected operations and what the semantics are. Norman: Simon, Brian and Amy should identify the collection of issues that they would like most to discuss at GGF12. To choose these and prepare a number of slides where each of them can do a couple of issues with a consistent slide sets. Have about 50 minutes, 10 mins intro/overview. Brian: for relational there are no open issues ... there are comments in the specs ... was going to suggest go through the comments to see if they can be consolidated into issues ... Norman: should probably go into Grid Forge. Brian: were you proposing to state the changes in going from GGF11 to GGF12? Norman: I was going to take that as given. These are all visible in Grid Forge and no one has commented so far ... Mario: there has been no notice that the new specs are out.. Norman: I or you can do something about this ... session 2 is on mapping which is chaired by Dave. Rather than re-run previous discussion we should try to look forward ... there is the ogsa naming discussion. Simon and Savas will be there ... Simon: put some thoughts about what we understand by the different approaches but it is difficult to predict what is going to happen ... Norman: what we might do is enumerate the options: 1. WSRF-only (what the specs do just now) i) to identify the initial resource to which request are submitted ii) to identify a derived resource 2. OGSA naming effort. What the impact this has. The names are: i) human readable ii) abstract name iii) addresses Simon: it's not a case as to whether you use abstract names but rather how the interfaces are designed to take that into account ... Savas: ... whether the interfaces assume there is information to specify what the resource is in the message. Norman: if there is additional addresses then what ... Simon: as a holding position we could define the interfaces assuming that this information is made available in the message or, as Sastry suggested, provide operations that remove it or associate it ... Norman: we can either talk about: a) the sorts of message that you get - a syntax led discussion or b) Can have a discussion led by the behaviour that you are trying to obtain. If you do the second you use the DAIS interfaces to connect to a resource that already has a name then you have one of two options (assuming that you have a mapping WSRF version and you have another name or a mechanism for creating names): Go through an additional naming scheme that maps to WSRF end-point or you have additional operations. ... Savas: the main difference is when you want to do things with multiple resources ... Norman: none of the DAIS interfaces refer to multiple resources ... Simon: people have talked about service groups and you could take the approach that Savas suggest but the scenario that Savas described is where the difference is ... in the end it all boils down to the message formats... Norman: but we could approach it from the assumption that if you had a pre-existing naming scheme ... Simon/Savas: that would make no difference whatsoever ... Savas: if you do not use an extra level of indirection you could still do it. If you don't want to expose the naming then you do not have indirection. Norman: what we probably want to do is that this is not something that DAIS will sort out and that DAIS will participate in a wider context but rather try to capture requirements from DAIS that can be passed on to the OGSA naming effort ... Simon: things that we have discovered are that we are right down in the weeds - things are similar until you send the message, there are differences when you deal with multiple resources ... Savas: there is a minor difference in the description of the metadata ... with WSRF you know that some information has to be included in the message as that is not described in the service. If you see the WSDL for that service it does not know that an EPR is expected ... Simon: ... the bottom line is that as far as DAIS is concerned we can jump either way ... we can talk about the things that we've learnt and demonstrate the different approaches but we are not sure how you make the decision of which mapping to adopt ... unless someone comes a way of supporting both then we have to chose or have two different mappings ... Norman: so maybe this is something that we can do at GGF ... Susan there was a subgroup in OASIS to discuss how you use these things? Susan: will check where they are at. They are doing what you are trying to do. Norman: will some of these people be at GGF? Susan: it looks like they will not. Norman: it would be good to invite them along ... Susan: will approach them all and see if someone can attend ... Norman: trying to have a discussion in the DAIS space and hopefully we'll have. Simon: are you planning to come along Savas? Savas: yes. Norman: Susan are you interested in getting involved with this session? Susan: yes, not necessarily to present but I can get people involved. Norman: will leave this session to Savas and Simon to put some material together. Dais session 3 was down in two parts: socialising the specs - have some background work but not much further ... the other session was in files but this is now going to be another separate session ... We'll possibly withdraw that session. Brian: firstly apologies, we thought DFDL was setting up a joint session but they did not so we got our own BOF. Norman: there are 3 other groups that have mentioned a relationship with DAIS. So Brian what would you like from DAIS? Brian: the charter we have at the moment is still felt by the data area co-chairs to be loose and woolly and not definite enough. Would appreciate comments on this. Also, we want to point out what activities are going on this area - OGSA-DAI and other implementations. We still have Martin Westhead's file realisations document sitting in the DAIS space. Norman: so you don't want someone from DAIS and say something? Brian: I think we have enough material but would welcome attendance. Norman:Others that have asked DAIS for some involvement are GSM - have a session where DAIS state what GSM is about from the DAIS perspective and vice-versa. Need to decide who is going to monitor what's happening in GSM. Then perhaps Dave or I should monitor this and prepare some slides. One of the infod session mentioned a relationship to DAIS Susan: we've been shifting our session. This is going to be the second session. The idea was to explore how the spec relates to DAIS ... if someone wanted to officially come along then that would be fine ... Norman: but is there a requirement for up front material? Susan: we have plenty to talk about ... whether we use this as a view to 3rd party delivery ... Norman: DAIS has tried to stop its functionality where the infod functionality starts ... Susan: the way the infod is going is that it will fill in that gap so it might be worth looking at this spec. Norman: so you are asking at the latest document with an eye to see if this is all joined up ... Susan: and I don't think it will ... want to do a re-fresh before GGF so you may want to hold off ... Norman: if you could flag up when the latest document becomes available then that would be useful ... anything else ... Susan: the OGSA data design activity ... Norman: there are lots of us going to be there ... Dave sent an agenda. There is a data access break out to see if we have been commissioned ... need to ask Dave. Discussion of the data design effort. Norman: usually have a DAIS get together at GGF to plan what is going to happen. Do not have people to discuss object database. AOB? Brian: the get together .... Norman: that requires looking at the timetable ... anyone there on Sunday? I may post something on the list then. -----