EDS Utility Processor
EDS Utility Processor
Any VSAM cluster such as the EPILOG datastore needs to be backed up and restored occasionally for archival purposes and to reorganize the cluster to improve performance and space use.
When you restore data to a cluster, you can reorganize the cluster. To do this, unload it to tape (or a temporary disk dataset) with IDCAMS REPRO and immediately copy it back into a new or reusable VSAM dataset. Reusable means that the dataset is either empty (for example, newly defined) or marked REUSE, in which case the previous contents of the cluster are overwritten.
If the EPILOG collector writes its records to SMF, you don’t need to back up the EDS. It can always be recreated from old SMF data. However, doing a simple IDCAMS REPRO from an EDS backup tape would probably be a lot easier than recovering data from SMF, because the data might be scattered across many magnetic tapes. For this reason, many sites do back up the EDS.
KEBUTIL utility processor
The drawback to using REPRO to perform these functions is that you must restore the entire dataset. The KEBUTIL utility processor lets you specify that only part of the data be reloaded to the EPILOG datastore. This lets you handle your periodic backup and reorganization in one neat package.
For example, suppose you plan to keep about one month’s worth of data in the EPILOG datastore. After the start of a new month, you would copy the entire EDS to a backup tape using REPRO (see step UNLOAD in Figure 41 on page 137). The KEBUTIL processor could then be used to copy the tape back into the EDS, excluding all the data from the previous month (see step RELOAD in Figure 41 on page 137). If you had defined the cluster as UNIQUE it could not be marked REUSE, and so an additional step would be required to delete and re-define the cluster before using KEBUTIL to reload it. This not only gives you a backup tape of the previous month’s data but also deletes the old data from the EDS and reorganizes it to reclaim the DASD space. thilev.TKANSAM(KEIIUTIL) contains a sample job similar to this.
This method of deleting old data from the database is much more efficient than the KEBMAINT PURGE command; you should only use PURGE to delete small quantities of data from the EDS.