Backing up data sets

With the DUMP command, you can dump DASD data to a sequential data set, which can be a generation in a generation data group (GDG), or as objects in an object storage cloud. The storage medium for the sequential data set can be tape or DASD. When the output resides on DASD, it may be a basic, large, or extended format data set. When the output resides in cloud storage, data is storage as a set of objects.

DFSMSdss can dump data sets both logically and physically. Data sets are located by searching either the catalog or the VTOC.

You can select data sets for dump processing based on data set names and numerous data attributes, as discussed in DFSMSdss filtering—choosing the data sets you want processed.

To perform incremental backups with DFSMSdss, you can filter with BY(DSCHA,EQ,1) to dump only data sets that have changed since the last dump was taken. If you also code the RESET keyword, DFSMSdss changes the data-set-changed indicator (DSCHA) after successfully dumping the data set. For more information about the RESET keyword, refer to Backup with concurrent copy.
Note:
  1. If you are using DFSMSdss on data sets that DFSMShsm is also backing up, you should not use the RESET keyword because it might cause confusion as to which backup is the most current.
  2. DFSMSdss does not permanently record the names of candidate volumes during dump processing.

The data-set-changed indicator and the last-referenced date (REFDT) are supported for VSAM and non-VSAM data sets.

Temporary data sets might be included in the data set list at the beginning of a DFSMSdss job. These data sets are created and deleted by other jobs that are running while DFSMSdss is running. Because they are temporary, these data sets can disappear before DFSMSdss finishes. DFSMSdss can issue a message informing the user what happened only at the time DFSMSdss tries to access the data sets. To hold all the data sets in a volume for the entire DFSMSdss execution, write an enqueue installation exit to enqueue the volume for the entire job.

When you create backups of data sets with the DUMP command, you can make multiple (up to 255) dump copies with a single DUMP command. This is done by specifying multiple ddnames on the OUTDDNAME parameter. To specify multiple ddnames on the OUTDDNAME parameter, you could code:

DUMP  -
    DATASET(INCLUDE(**)  -
            BY(MGMTCLAS,EQ,MCNAME1))  -
    OUTDDNAME(TAPE1,TAPE2,TAPE3)  -
    COMPRESS

This technique can be helpful if you want to create several backup copies to be used for different purposes.

Unless overridden by the installation options exit routine, DFSMSdss continues dumping while at least one output copy does not have an output error. In the event of an abend, however, DFSMSdss ends without completing any backups.

For more information about the data-set-changed indicator and REFDT, refer to z/OS DFSMS Installation Exits.