Data set level recovery performance considerations

The elapsed time for data set level recovery varies significantly based on the recovery method that is used.

• FlashCopy from disk. When FlashCopy is used to perform the recovery, the recovered data set is available for use in a matter of seconds. Note that FlashCopy is not the default method for performing a recovery, because the recovery creates a FlashCopy relationship between the backup volume and the production volume. This relationship prevents a backup of the copy pool from being created until the background copy has completed and the relationship is terminated. The SETSYS FASTREPLICATION(DATASETRECOVERY(option)) command can be used to change the default.
• Standard I/O from disk. When the recovery is from disk, standard I/O is the default recovery technique. The elapsed time for the recovery varies depending on the size of the data set. Small data sets may complete in seconds, while larger data sets may take minutes to tens of minutes.
• Standard I/O from tape. When the recovery is from tape, the elapsed time is significantly longer. The increase in the length of time is due to the time that it takes to find the data set on tape. Since the backup copies are created as a full volume dump, DFSMShsm does not record the file block ID of each data set. This results in DFSMShsm having to read through the entire dump volume contents to find the data set before it can begin recovering the data set. The time of the search is longer for larger volumes. This means that DB2 object level recovery from tape takes significantly longer than a recovery from disk. It also takes significantly longer than an object level recovery from a DB2 image copy created on tape.