DAIS Session 4: Data Distribution and Charter Discussion -------------------------------------------------------- GGF10, 11th March 2004 Chair: Dave Pearson Notes by Mario Antonioletti (mario@epcc.ed.ac.uk) Actions Derived from this session: [action] Need to go and talk to the people that contributed to that scenarios document. Expected the scenarios to be described but not how they would be supported. As the scenarios are changing it would be a good idea to express the scenarios without an actual mapping.. [action] See if there is more interest on object databases in the dais mailing list. [action] Solicit further information on Data Accesss Request Composition to the mailing list before committing to having this as a deliverable. +----------------------------------------------------------- Presentations available from: https://forge.gridforum.org/projects/dais-wg Documents->GGF Documents/GGF10/Presentations Make sure that the document filter is working correctly. Unattributed statements in general belong to the speaker at that point in time. +----------------------------------------------------------- Dave Pearson introduced the session. [See the Intro_to_the_third_DAIS_Session presentation on Grid Forge Note - that there is a certain level of inconsistency in the session numbering as these were changed in the point before/during GGF and the two labelling schemes were used by folks.] Going to cover two topics: - Data Distribution: Chris Kantarjiev - Charter Discussion: Norman Paton There was a information dissemination. Covered some interaction patterns that are being used by DAIS and those that will be covered by information dissemination. ---- Chris Kantarjiev presents the Information Dissemination (previously called Data Distribution) work. [See the Information_Dissemination presentation on Grid Forge]. Presented Data Distribution at GGF9 - as was called then. Collected issues then and taking these into account the document was updated. Have not made any effort to refactor the document in terms of WSRF. ... Want to support existing data protocols and not invent our own. Information dissemination will just be another port type at the same level as the other DAIS portType. Determine how data is going to be published. How to do subscription and how the data is propagated. Propagation is analogouse to WSRF model but it covers different things. There is an interface for conumption which allows for extra filtering (other interfaces mentioned in the slides). All the activities are specified by rules. Rules are divided into four different components: - what - when - clock time - changes in the underlying data set - a state change occurs in the database - where - QoX Norman: in some instances though you may not want to be notified of an event if the change takes place within a transaction ... (if the account goes below 5000 and over again in the same transaction and you have asked to be notified if the account goes below 5000 you would not want to be notified of this temporary change). Issues raised in DAIS f2f ID portType - Access or Data Management Covered some other issues covered in the f2f. Presentation followed the text in the slides fairly closely. Asked for help in the transactional context. Dave Pearson: worth considering that there has been a transaction BOF and that DAIS together with Data Dissemination will cover this with this new RG. Transformations would be performed as a stored procedure. Dave: the point of transformations is important to us, we have discussed it in DAIS. The data that you want to consume may not be in the right format which is why data is important. Cover scenarios that are not covered in DAIS. We have not seen the extensions that Greg Riccardi was going to cover. Went over some of the scenarios in more detail: scenario 2, 3a, 3b, 3c and how they would be satisfied by Information Dissemination. (these were presented by Dave in the introduction to the session). What's next: - do more work on the specification of the rules - produce a working group which will take us in new directions. Dave: it would be useful to solicit for use cases. David Martin: do you see all of this being done ....? If there are two groups. Chris: do not want to build the stuff that DAIS is going to build. Our focus is on publishing, subscription and publication and the things that go along with that. Dieter Gawlick: want to make independent as possible and do not want to write our own model ... Peter Kuntz: what kind of assurance do you get if you register as a consumer? Chris: did not mention the QoX that we have been covering ... Dieter: can depend on the work that has been done on the database area and use that knowledge. Chris: there are other places where the description of the work is being done ... WS-Agreement ... Norman: particle physics applications are disseminating information but do you see (directed to Peter) things in the way that they have been presented ... Peter: it depends what you want to do ... this stuff is very low level ... this could be good for certain things ... Dieter: this is a superset of the semantics of JMS. Peter: ... we can help by trying to come up with use cases.. Dieter: are you going to be a volunteer for use cases ... Peter: possibly.... ---- Norman talks about Chaters and Plans. [See the DAIS-WG_Charter_and_Plans presentation on Grid Forge]. The purpose of this part of this session is to discuss deliverables and timescales. Group has been in existence in less than 2 years. Our objectives have changed with the top level goals being the same but the time scales changing. New groups formed and underlying changes in the specs we rely on. We are hoping by june 2004 most of our specs will be nailed dwon: - remove ogsi - remove ubiquity of mapping materials - interim position on actual mappings When do we hope to have our recommendations out then? March 2005? Difficulty to say as there are dependency on other recommendations. David Martin: you mean on WS-*? Yes and other dependencies. David Martin: What the SG has been discussing is to encourage groups to do the best they can and set the dates and realise that things might slip. Norman: would you allow recommendations to be submitted that have external dependencies? David: that is a good question. I do not know but you should try to get things as ready as possible. Some people are freezing their documents and proposing to put them through as informational documents. Scope issues: - reduction of scope - informaiton dissemination has moved elsewhere. - Increase in scope - explicit inclusion of file access - would need to hold a BOF Have not achieved a critical mass. Need volunteers to be involved in this. Is there anyone that might be interested in leading this? *silence* We either need to establish a team or we can explicitly stack it - what is the process? David: not too bad to continue hosting the discussion but don't have a milestone for this. Simon Laws: it's still useful to have files in the discussion ... would not want to move it out of scope. Susan Malaika: but GFS could take it over ... David: but the scope there is different. Peter: in our storage management BOF .... Norman: we were not very clear on the scope of file access on DAIS Peter: in the storage management BOF this is seen as another mechanism to access the file - so the assummption .... Norman: in DAIS we are fairly agnostic about the underlying data resource ... access could mean a variety of things ... if your route gets approval then you would have a route to operate on things that we do not define as access, i.e. does not access the file, then you can extend the scope to areas that we might call management ... sounds ok ... Dave Pearson: the data area workshop discussed the value of having an architecture ... if you got people working on an architecture it would resolve those kind of arguments. David: I agree with that .... Norman: in OGSA the data section is less filled out than it might be it's going to be addressed after GGF11. This will lead to different emails being sent out to different people to categorizes the different services.. Peter: this is a good way to proceed ... digestable chunks that a community can standardise. David: do not want slow any group down ... I understand Norman's issue about not creating problems for the future .... Another extension point is the explicit inclusion of object databases - a couple of people indicated that they might be interested in doing this. Two people put their hands up: - Dr Ammar Benabdelkader - Dr Thomas Soddermann We can then publish to the list and see if there are some more people. We'll take an action to see if there is broader interest. In terms of the group there was a compositional section in previous version but we backed down - hope that this will go through as a community practice document. Deliverables: - requirements and functionalities (available GFD.13) - data service scenarios (capture scenarios - due June 2004) expected the scenarios to be described but not how they would be supported. As the scenarios are changing it would be a good idea to express this. Simon: forgotten that my name appeared in that document ... need to take a look at it. - representing data services using web services. This was covered in the first session. It may turn out to be an informational document ... although June might be too early for that. Simon and Sastry are ok about taking this document forward as an informational document ... this might be October. Norman: might have a workshop at the next GGF on these mappings issue. (show of hands was requested - all that expressed an opinion were positive - i.e. no one said no) Next set of deliverables have dependencies on wsrf - OGSA Data Services indication is to wait to see how things turn out. David Martin was not scared by the number of the deliverables. Mode of operation: Have had one face-to-face meetings between GGFs. Do not have a proposal for a full f2f meeting before ggf11. GGF12 is going to be in the latter part of September. Not good for academics as that's when college starts. There will be one small meeting for spec authors. Telcons will continue weekly. Main authors have been stable for some time but we welcome contributions from anyone out there - make yourself known if you want to participate more closely. Norman encouraged folks to go to the CIM WG meeting. Session was drawn to a close.