Summary of data management meeting Tuesday, February 24th, 2004 9:30 Bat 2-R030
===============================================================================
         
Present: N.Brook, P.Charpentier, J.Closier, C.Cioffi, U.Egede,
         M.Frank, R.Graciani, G.Kutznetsov, J.Saborido, 
         A.Tsaragorotzev, R.Vizcaya



File catalog (Andrei):
 The AliEn file catalog is back to normal conditions and running in 
 Marseille. It will be installed on a dedicated machine by Joel.
 Andrei has understood the information in the POOL XML catalog and
 knowns how to populate the Alien catalog. Files will be registered
 with the first "stable" replica availible. The physical file name
 will contain the site, where the replica is stored and the protocol
 of access.
 In order to access files across GRID boundaries, it will be necessary
 to store files with an already assigned GUID. Otherwise files created
 on the LCG test bed will not be accessible outside LCG (e.g. in normal
 batch processing) and vice versa. This must be investigated.
 This means that the GUID is a mandatory item of any file, which
 should be selected using the bookkeping database.

 In the aftermath of the crash of the Alien file catalog several weak
 points were discovered:
 - Alien is not a production system. It is a working prototype and
   gaining experience is mandatory for successfule operation. One
   of the problems was a missing file, which was not backed up.
 - Alien is running on mySQL. Backups must be done by us.
 - We cannot rely entirely on the Alien file catalog. We need the
   slices into the POOL xml catalog.
 - For this year's data challenge we will only rely on the existing
   bookkeeping database. The Alien catalog may be optional. The 
   bookkeeping database will have to provide an interface which
   can be used by the DIRAC match-maker.

The high importance of the POOL XML catalog triggered a very long
discussion. Here is a short summary:
- Slicing any file catalog and producing the XML file needed
  by POOL is mandatory. Philippe required that any physicist must
  be able to create with simple tools such a slice given a set of
  input files obtained from the bookkeeping database.

  Action: Carmine and Markus will investigate the possibility
  of creating slices from the existing bookkeeping database.

The file catalog issues were discussed at length and took up 
all remaining time of the meeting.



Minutes:

     Markus Frank