Improving recovery performance

You can improve your recovery performance by taking more frequent image copies.

About this task

You might want to limit this activity by determining which table spaces most need fast recovery. The following guidelines are provided as a starting point to help you determine how often you must do incremental image copies. As with a single subsystem, doing frequent image copies can help you avoid using the archive log for recovery.

Procedure

To improve recovery performance:

Use the following guidelines to increase the frequency of image copies for each member of the data sharing group. Use the output of the DSNJU004 (print log map) utility for each member.
  1. Find the starting timestamp of the active log data set with the lowest STARTRBA.
  2. Find the ending timestamp of the active log data set with the highest ENDRBA.
  3. Calculate the time interval: 
    time_interval = end_TIMESTAMP - start_TIMESTAMP
  4. Calculate the interval at which to perform incremental image copies:
    interval of copy = time_interval * (n-1) / n

    n is the number of active log data sets.

  5. Take the smallest interval for the group and, to account for fluctuations of log activity, decrease the interval by 30%. (30% is an arbitrary figure; you might have to adjust this interval based on your system's workload.)

    This is the recommended interval for doing incremental image copies. If the interval is too small to be realistically accomplished, consider increasing the size or number of active log data sets.

    Periodically run the MERGECOPY utility with incremental image copies. The RECOVER utility attempts to mount tape drives for all the incremental image copies at the same time. If it runs out of tape drives, it switches to log apply. MERGECOPY merges what it can and then mounts more incremental image copies.