Parallel Recovery
Parallel recovery can speed up both instance recovery and media recovery. In parallel recovery, multiple parallel slave processes are used to perform recovery operations. The SMON background process reads the redo log files, and the parallel slave processes apply the changes to the datafiles. Parallel recovery is most beneficial when several datafiles on different disks are being recovered.
In a serial recovery scenario, the SMON background process both reads the redo log files and applies the changes to the datafiles. This may take a considerably long time when multiple datafiles need to be recovered. However, when parallel recovery is being used, the SMON process is responsible only for reading the redo log files. The changes are applied to the datafiles by multiple parallel slave processes, thereby reducing the recovery time.
Recovery requires that the changes be applied to the datafiles in exactly the same order in which they occurred. This is achieved by single-threading the read phase of the recovery process by the SMON process. SMON reads the redo log files and serializes the changes before dispatching them to the parallel slave processes. The parallel slave processes then apply those changes to the datafiles in the proper order. Therefore, the reading of the redo log files is performed serially even during a parallel recovery operation.
The RECOVERY_PARALLELISM initialization parameter controls the degree of parallelism to use for a recovery. You can override ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access