Parallel recovery can be accomplished in two ways.

Option 1   Use several SQL*Plus sessions to do recovery.

For example, if you had three tablespaces to recover, you could recover each tablespace in a different session simultaneously. Although this might be better than performing a single process recovery, all processes must still read all the log files used in recovery.

Option 2   Use the PARALLEL option of the RECOVER command.

This method works just like Oracle's Parallel Query Option. From one SQL*Plus process, Oracle reads the redo log file and dispatches redo information to multiple recovery processes that will in turn apply the changes to the datafiles. The number of recovery processes is determined by the degree specified in the RECOVER command.

As with parallel query, parallel recovery benefits most when the datafiles being recovered are on many different disks. Because recovery is very disk intensive, no more than two recovery processes per disk should be used.