Joint discussion with DAL,DM,Registry,VOQL,Semantics about SimDB on Wednesday 29 Oct 2008 8:30-10:00, room B475

Presentation by HerveWozniak available here

Minutes of the meeting (taken by ClaudioGheller)

Herve presentation:

- Goal: find agreement around SimDB. Address correctly (compatible to other VO standards) the problem.

- Review what data we deal with. Differences between theoretical and observational data. Th. data are more complex. Data last not too much. However old data could be used to be "confirmed" with new results and methods. The turn-over is however fast. Timelife is an issue.

- Data access to data producers and to external users as well.

- The general feature of Th data is diversity (in all aspects): services to access and discover data needed.

- dedicated standard required to access data. Queries are very different from Obs. (and probably more complex).

- SimDM: data model + query service to the derived database. Describes: * the protocol (software) * the experiment (simulation) * snapshot * minimum information on resources to handle the data Finer than registry Different groups are interested VOQL, DM, REGISTRY , DAL, SEMANTICS

- What does TIG need? * Input from WG on SimDB NOW (in designing phase) * Ensure similarity with, reuse of other VO works * Need representative of othere WG (Mireille Louys-DM, Norman Gray-Semantics: OK)

Need a core team around SimDM.

Discussion

Rick: the data model is more or less OK. It is the interfaces that requires work and contributions.

Roy: howledge w make Th. services compatible with standard VO services? In particular what's harvesting for?

Gerard: a tool can go to the registry and find the data that is suitable.

Pedro: why not using what is already developed (TAP+ADQL) posing the questions in a different way? No new way to solve possible problems, but fine tune to what is already there. Stress that tools are there.

Genova: include the present knowledge i Th work is needed.

Pedro: the direction is the opposit. Not the WG that enters the TIG, but the TIG that asks to WGs of interest

Herve: formal representation of WGs is NOT required. Just help from experts.

Pedro: confirms that this is the good direction.

Roy: lack of knowledge by TIG on how things proceed in the VO. First, try to fill the gap. The SimDAP is fundamental to have a common description of data for searching and discovery. The standard search model can be applied to Th. What's the role of the registry for Th.? Do data centre can manage also other's data publishing it, registering in some way this service? The same for processing services.

Rick: yes. The hosting is on metadata and not on data.

Herve: data cannot be moved. Only metadata can.

Genova: confusing the description of the data and the interoperability issues.

Norman/Rick: a standard vocabulary is needed for SimDB.

Pedro: stress again that the direction should be: representative of Th. goes to WGs to solve the problems.

Herve: is this efficient?

Pedro: the opposite model is not sustainable

Herve: however the procedure could become too slow

Pedro: following the standard procedure is the best way

Genova: there is only ONE TIG (not many...), so the effort should not be too much for WGs. So help TIG people to enter the VO. Probably TIG people cannot pose the correct questions and so some direct support form WGs wolud be necessary, until the TIG is completely "IN" the VO.

Herve: confirm that a more dedicated effort is necessary

Roy: start simple. Don't think of SimDM as a registry.

Igor: which relation between different data models?

Herve: still to be discussed.

CONCLUSION: All interested WP leaders agree in helping TIG with a representative. Direct involvement of WGs in TIG.

Add-on by LaurentBourges

The main idea for SimDB is to reuse the TAP / ADQL (asynchronous mode) work to query the database. According to Pedro Osuna, it is possible to do that and reuse existing frameworks. Issues can be raised to the DAL WG for TAP or VOQL WG for ADQL concerns.

Ray Plante proposed to concentrate on the query service and use the registry to discover SimDB services (TAP query interface) and then those TAP interface will be used to run complex queries on the SimDB data model.

Francoise explained that TIG & SimDB involved people need help from other groups (DAL, REG) to learn what is already present in the VO. Of course, we can raise issues to WGs but we need a better understanding of VO standards (Registry, TAP for example).

-- HerveWozniak - 29 Oct 2008