RESTORE/RECOVER › Restore and Recover of VSAM Clusters › Restoring Alternate Indexes which were Archived Separately

Restoring Alternate Indexes which were Archived Separately

In addition to the methods discussed previous, alternate indexes can also be archived and restored separately from their base clusters. For example:

Archive an alternate index by itself (CA Disk will retain the association information from the catalog in the FILES):

FIND DSN=alternateindexname
ARCHIVE DISP=RECAT

Restore an alternate index by itself and re-establish association with the base cluster:

RESTORE DSN=alternateindexname

Note: In order to restore an alternate index as discussed previous, the associated base cluster must be preallocated and cataloged in the same catalog as the alternate index.

There are three critical precautions you must follow if you want to separately archive and restore alternate indexes:

• In archive, since the base cluster is not being archived at the same time as the alternate index, it is imperative that no updates to either file occur after the first archival begins. CA Disk cannot ensure this, so you must control this manually in order to keep the files synchronized. This becomes complicated if the data sets are selected implicitly by DSCL, because CA Disk processes the data sets in ascending data set name sequence without regard to the associations between the data sets (that is, the base cluster/alternate index associations). Finally, if you are going to archive the alternate indexes separately, the best thing to do is explicitly archive the base cluster and all alternate indexes in a single archive job. For example:

  FIND DSN=alternateindex1name
  ARCHIVE DISP=RECAT
  FIND DSN=alternateindex2name
  ARCHIVE DISP=RECAT
  FIND DSN=alternateindex3name
  ARCHIVE DISP=RECAT
  FIND DSN=clustername
  ARCHIVE DISP=RECAT
  

• In restore, since the base cluster is not being restored at the same time as the alternate index, it is imperative that no updates to either file occur until after the last restore completes. For example, if the base cluster is restored first and the alternate index second, if the base cluster is updated prior to the completion of the restore of the alternate index, the two files will be unsynchronized. CA Disk cannot detect this type of error and no error will be indicated. After a base cluster and alternate index become unsynchronized, any application that accesses the base cluster through the alternate index can return erroneous information without the user being aware of any problem.
• Another problem can develop if you attempt to separately restore an alternate index that was defined over an ESDS cluster. If you restore the base cluster with a different CISIZE than its defined original, your separately restored alternate index will probably be invalid. This occurs because the alternate index contains pointers to the base cluster based on the relative byte address (RBA). When you restore the base cluster with a different CISIZE, the relative addresses of the records in the base cluster will probably change after the first control interval, thereby making all other offsets in the alternate index invalid. Although you can restore this alternate index, it cannot be used to access the data in the base cluster.