DFSMShsm backup functions that exist for non-VSAM data sets also
exist for VSAM data sets cataloged in the integrated catalog facility
catalog. DFSMShsm backs up VSAM data sets on level 0 volumes by using
the Access Method Services EXPORT command or DFSMSdss as a data mover.
The backup of a VSAM data set is performed only on a complete VSAM
sphere. The base cluster and all alternate index clusters defined
on the base cluster are backed up as part of the data set and are
not specified separately. Paths are included automatically in the
data set backup.
Note: When a VSAM compressed data set is
backed up, DFSMShsm suspends any software compaction requested and
records in the BCDS that the data set is striped, compressed, or both.
DFSMShsm can back up multiple-volume, SMS-managed, VSAM data sets
when it processes the first volume of the base data component. Normally,
DFSMShsm prevents update access to any VSAM data set that is currently
being backed up. This is done to prevent possible inconsistency of
the backup copy.
Tip: You can override this protection
for VSAM data sets with the SETSYS INUSE RETRY command or the ARCBDEXT
exit.
Some installations, however, would prefer that their jobs not fail
when they attempt to open a VSAM data set that is currently being
backed up. There are two methods to accomplish this, one which includes
the risk of having a corrupted backup copy, and one which excludes
that risk:
- One method is to use the DFSMShsm data set
backup installation exit, ARCBDEXT. This exit can be coded to change
the way DFSMShsm performs serialization on VSAM data sets, allowing
batch jobs to open the data sets while the data sets are being backed
up. This technique, however, exposes installations to the possibility
of creating unserialized backup copies.
- The other method is to add an IEFBR14 step to any batch job that
accesses any VSAM data set for updating. The only function of the
added step is to access the data set with a DISP of MOD or OLD. This
step should be placed immediately before the step that accesses the
data set for updating. The IEFBR14 step attempts to get an exclusive
ENQ on the data set, which causes the job to wait until DFSMShsm has
finished processing the data set before the subsequent step accesses
the data set. This method may require more time to implement than
the ARCBDEXT solution, but it does eliminate the possibility of having
corrupted backup copies.