Managing Backup and Recovery

Learn how to use Recovery Manager (RMAN) to back up and restore Oracle Real Application Clusters (Oracle RAC) databases, and about Oracle RAC instance recovery, parallel backup, recovery with SQL*Plus, and using the Fast Recovery Area in Oracle RAC.

Managing Backup and Recovery in Clusters

In an Oracle RAC cluster, any node can restore archived redo logs because data files are accessible from every cluster node; the instance performing recovery does not need to restore all data files itself.

RMAN Backup Scenario for Noncluster File System Backups

In a noncluster file system configuration, each node backs up only to locally mounted storage and cannot access archived redo logs on other nodes unless network file system access is configured. With shared network access, each node still backs up its own archived redo logs to its local directory.

RMAN Restore Scenarios for Oracle RAC

Oracle RAC restore procedures depend on the storage architecture: cluster file system backups can be restored from any available node with access to the shared files or media, while noncluster file system recovery requires network access to every node’s archived-log directory. RMAN can restore the SPFILE to its default or a specified location, and Oracle Enterprise Manager can restore it when the database is closed; channel allocation may need to be adjusted for load balancing during full restores.

Instance Recovery in Oracle RAC

Oracle RAC automatically performs instance recovery after software or hardware failures by applying online redo logs, preserving committed transactions, rolling back incomplete work, and releasing resources. Oracle Clusterware restarts failed instances, surviving instances recover failed peers, and if all instances fail, the next instance to open the database performs recovery; all online redo logs must be accessible, so mirroring is recommended. RMAN is the standard tool for backing up and restoring RAC data files, control files, SPFILEs, and archived redo logs.

Media Recovery in Oracle RAC

Media recovery in Oracle RAC is initiated by a user through a client application, typically using RMAN to restore data file backups and recover the database. The recovery node must access all required data files and archived redo logs, either on disk or through backups; for encrypted tablespaces, the Oracle Wallet must be opened after mounting the database and before opening it.

Parallel Recovery in Oracle RAC

Oracle AI Database automatically selects an appropriate recovery parallelism based on available CPU resources. With RMAN, the number of allocated channels controls parallelism for restoring data files and applying incremental backups, while archived redo log application is automatically parallelized; parallel recovery can be disabled through initialization parameters or the RMAN and SQL recovery commands.

Using a Fast Recovery Area in Oracle RAC

An Oracle RAC fast recovery area must reside on shared storage, such as an Oracle ASM disk group, cluster file system, or network file system directory accessible by every instance. The DB_RECOVERY_FILE_DEST parameter must have the same value across all instances, and Oracle Enterprise Manager provides configuration options through the cluster database Maintenance and Backup/Recovery settings.