Using BACKOUT for point-in-time recovery without image copies

Only the log between the point in time specified by TOLOGPOINT and the current log point is read and processed with this technique, which may substantially reduce log processing. In addition, the spaces are read for log backout processing, and image copy processing is completely eliminated. If indexes are not rebuilt and output image copies are not requested, only the pages updated by log records are read and written, which can greatly reduce I/Os and speed up the recovery. (For more information, see Using copies and logs for index recovery.)

To convert an existing partial recovery job to this technique, simply add the BACKOUT YES option to the usual syntax. Add the INDEXLOG YES or INDEXLOG AUTO option and BMC AMI Recover assumes that the related indexes should be backed out to the same point in time.

OPTIONS BACKOUT YES INDEXLOG YES
RECOVER TABLESPACE PAYROLL.EMPLOYEE
        TABLESPACE PAYROLL.RULES
        TOLOGPOINT LASTQUIESCE
RECOVER INDEX (ALL) TABLESPACE PAYROLL.EMPLOYEE
RECOVER INDEX (ALL) TABLESPACE PAYROLL.RULES

When you select this strategy for potential point-in-time recovery, image copies may be avoided prior to batch runs without risking lengthy recovery times. Image copies do not need to be made of indexes to use the BACKOUT technique. (Indexes would still need to be rebuilt for media failure or disaster recovery.) COPY NO indexes are eligible for BACKOUT processing.

After a backout recovery, you must make a copy of the index before a recovery to current (by using a copy and log records) can be done. To maintain the ability to recover from log in case of a future media error, you can start an BMC AMI Copy SHRLEVEL CHANGE copy running in parallel with other applications, as soon as the backout recovery is complete.

You can use the OUTCOPY YES option in your BACKOUT run. However, this request may degrade performance, because making an output copy causes all pages of the index space to be read.