Administering Storage in Oracle RAC

Oracle recommends Oracle Automatic Storage Management (Oracle ASM) as a storage management solution that provides an alternative to conventional volume managers, file systems, and raw devices.

About Oracle ASM

Oracle Automatic Storage Management (Oracle ASM) is a volume manager and file system for Oracle AI Database files in single-instance and Oracle RAC environments. It organizes storage into disk groups, distributes file contents evenly for consistent performance, supports online disk addition and removal with automatic rebalancing, and provides normal, high, or external redundancy options. Together with Oracle Managed Files, ASM simplifies file creation, naming, deletion, and storage administration while supporting coexistence with raw disks and third-party file systems and management through SQL*Plus, ASMCMD, and ASMCA.

Overview of Storage Management for Oracle RAC

Oracle RAC requires all data files, instance undo tablespaces, and redo logs to reside on shared storage accessible by every instance. Oracle recommends Oracle ASM disk groups, although certified cluster file systems are also supported, along with a shared SPFILE containing instance-specific settings. Raw block storage is no longer supported by Oracle AI Database or Oracle Clusterware, while storage features such as Oracle ASM, Oracle Managed Files, and automatic segment-space management generally operate in RAC as they do in single-instance databases.

Data File Access in Oracle RAC

Every Oracle RAC instance must access every data file, and instances verify file accessibility during startup and when tablespaces or data files are added or brought online. Verification fails if a file is placed on storage unavailable to an instance or if instances access different copies of the same file. After correcting access problems, run ALTER SYSTEM CHECK DATAFILES on each instance to confirm consistent data file access.

NFS Server for Storage

Oracle AI Database can operate as an NFS server, responding to client requests while storing files and metadata inside the database. It can automatically replicate files associated with a primary database to a standby database and can store unstructured content such as email. Oracle file systems can be created or destroyed and accessed through the NFS server using the documented SecureFiles and Large Objects procedures.

Redo Log File Storage in Oracle RAC

Each Oracle RAC instance must have at least two redo log groups, which Oracle DBCA allocates automatically during database creation; group counts and sizes can be adjusted later, and adding a cluster node configures its redo logs. Instances switch groups as logs fill, archive completed groups when running in ARCHIVELOG mode, and participate in recovery checks. When an instance is removed, its redo thread should be disabled so recovery does not unnecessarily examine it.

Automatic Undo Management in Oracle RAC

Oracle AI Database automatically manages undo segments in an undo tablespace assigned to each RAC instance, while all instances can read cluster-wide undo blocks for consistent reads and can update eligible undo tablespaces during recovery. Assign each instance through its UNDO_TABLESPACE parameter in the SPFILE or PFILE, with automatic allocation available when Oracle Managed Files is enabled. All RAC instances must use the same undo management mode, and policy-managed database deployment is desupported in Oracle AI Database 26ai.

Oracle Automatic Storage Management with Oracle RAC

Oracle ASM distributes database files evenly across disk-group storage, can mirror files for protection, and reduces storage administration without requiring I/O tuning in RAC. Grid Infrastructure creates an ASM instance on each cluster node, with each instance using an SPFILE or PFILE; parameter files and nondefault listener entries should be backed up. ASM disk groups require shared, properly permissioned storage and can be administered with ASMCA, Enterprise Manager, SQL*Plus, SRVCTL, or installation tools, while CVU verifies ASM and shared-storage integrity across nodes. ASM automatically rebalances files after disk changes, coordinates metadata among instances, and supports preferred read failure groups for locality in extended-distance clusters through ASM_PREFERRED_READ_FAILURE_GROUPS. Installing Grid Infrastructure converts nonclustered ASM to clustered ASM, and SRVCTL can add, remove, start, stop, configure, and report ASM instances. Disk groups can be created and managed independently of database creation through ASMCA or Enterprise Manager.