Summary of data management meeting Tuesday, January 13th, 2004 9:30 Bat 2-R008
==============================================================================

Present: N.Brook, P.Charpentier, C.Cioffi, J.Closier, M.Frank, V.Garonne, 
         R.Graciani, G.Kutznetsov, F.Loverre, I.Stokes-Rees, A.Tsaragorotzev


Status reports:

Workload Management System (Andrei, Ian, Vincent):
- All required functionality is implemented with the exception of the
  file catalogue. Andrei will start to implement the usage of the file 
  catalogue this week.
  Tests using the LCG 2 environment did not yet start. It is expected,
  that functional installations will appear in the course of the next
  two weeks. Tests will only start once they are availible.
- Files produced outside LCG will be accessible from LCG and vice versa
  (confimed by Nick later by e-mail). All files must be registered
  by the corresponding file catalogues.
- Storage elements currently are implemented using bbftp. It was 
  mentioned, that bbftp will have to interact with the local mass storage
  and the local file caches.
- Work on the monitoring API is ongoing and Ricardo presented a
  first solution for a display. This display summarizes the information in
  the job database.
- Stress tests show, that there are no obvious bottle necks.

File catalog (Andrei):
- An instance of the Alien file catalog is installed in Marseille.
  It was mentioned, that outbound networking connections are essential.
- A python API is availible. 
- The C++ client is being worked on by the Alien team.

  Action: Andrei will interface Dirac to the Alien catalog

File merging discussion (All):
- Please see seperate documents.

Dirac desktop (Gennady):
- It is possible to instantiate jobs using the work flow editor.
- The connection of the work flow editor to edit the data production
  database has to be done.

  Action: Gennady will implement the connection to the production database.

Pool integration (Markus):
- The LHCb event model is persistable. The problems saving the 
  last remaining class (2D weighted relations), which cannot
  be saved using the current release should disappear with the next
  release of POOL.
- The performance problems were strongly reduced.
- The usage of the POOL XML file catalogue will be problematic.
  Another possible solution would be to implement the
  POOL catalogue interface using the Alien catalogue.

  Action: Markus will investigate possible solutions.

Hardware:
- It was agreed, that 2 extra machines, which could host
  dirac components should be on stock for the data challenge.

  Action: Joel will organize the necessary hardware.


Cheers,

     Markus Frank