Reliability of the file catalogue

It is clear that the file catalogue must be extremely reliable, if it is to be used for access to thousands, perhaps even hundreds of thousands, of datasets. To this end, once an entry has been added to the file catalogue it is guaranteed. This is performed using various levels of journaling, described below.

  1. Updates to the file catalogue consist of separate files, in FZ format. If these updates cannot be successfully processed, they are stored in a special directory (or mini-disk), for subsequent automatic retry.
  2. Once updates have been successfully added to the local RZ database, they are then sent to all other systems as indicated in a distribution list. Again, automatic retry is provided.
  3. Once an update has been successfully added to the local database and sent to all remote servers, it is moved to a journal area. These journal files are purged regularly, based upon the interval between regular disk backups.
  4. On systems where the SQL interface is enabled, updates are automatically added to both RZ and SQL databases.
  5. In addition, the RZ databases will normally be backed up on a regular basis. This, together with the local journal files, allows a remote site to reconstruct its RZ database.