This module introduce the recovery-catalog maintenance, reporting, automation, and cloud-backup topics covered as well as RMAN recovery catalog and cloud backups in Oracle 26ai. Oracle Recovery Manager (RMAN) records the metadata needed to identify backups, copies, archived redo logs, database incarnations, and persistent configuration. The target database control file is always an RMAN repository. A recovery catalog is an optional database schema that stores resynchronized RMAN metadata for one or more registered target databases.
Module 4 explained how to connect RMAN to the target database and recovery catalog, work in NOCATALOG mode, select an incarnation,
resynchronize metadata, and maintain repository records. Module 5 builds on that foundation. It moves from basic repository maintenance to adding
existing backup files, protecting or reconstructing catalog information, evaluating recoverability, automating work, and reporting across registered
databases.
This module targets Oracle AI Database 26ai. Recovery catalogs, stored scripts, and the familiar RMAN reporting commands remain current, but the operational context has changed substantially since Oracle8 and Oracle Database 11g. Current examples must account for multitenant databases, Data Guard environments, dedicated administrative privileges, longer metadata histories, provider-specific cloud modules, and restore testing.
Repository metadata and backup data are different layers. The control file and recovery catalog describe backup objects and database history. Backup storage contains the backup sets, backup pieces, image copies, and archived redo logs that RMAN may need during a restore or recovery.
Target control file -> metadata for its target database
Recovery catalog -> metadata history for registered target databases
Backup storage -> backup pieces, image copies, and archived redo logs
A repository record does not prove that its associated file still exists, can be read, or can satisfy a recovery objective. LIST and
REPORT describe recorded metadata. CROSSCHECK reconciles repository status with files that RMAN can access through the
configured device. VALIDATE reads data or backup content for documented validation tasks. A tested restore remains the strongest
operational evidence that the backup strategy can recover the database.
The recovery catalog extends the control-file repository rather than replacing it. Organizations commonly use a catalog when they need:
The catalog is metadata, not a second copy of the backed-up database. It requires its own availability, backup, recovery, access-control, and monitoring strategy. The catalog must not reside in the target database that it is intended to help recover, and the catalog database should not share the same failure domain as protected targets when that can be avoided.
The lessons follow the path from locating backup metadata to using that information safely:
| Lesson | Topic | Operational outcome |
|---|---|---|
| 2 | Adding existing backup metadata | Use CATALOG to inspect supported existing files and record them in the RMAN repository. |
| 3 | Protecting and rebuilding catalog information | Recover the catalog normally when possible, or reconstruct the metadata that remains available. |
| 4 | Analyzing repository metadata | Use REPORT to examine schema history, backup need, obsolescence, and unrecoverable changes. |
| 5 | Displaying recorded objects | Use LIST to display backups, copies, archived logs, incarnations, and script information. |
| 6 | Creating stored RMAN scripts | Create and manage local or global scripts in the recovery catalog. |
| 7 | Executing repeatable RMAN work | Use RUN blocks and execute stored scripts with explicit monitoring and error handling. |
| 8 | Querying recovery catalog views | Use documented RC_ views with the correct target database and incarnation scope. |
| 9 | Operational review | Combine catalog maintenance, reporting, automation, and verification into a controlled workflow. |
The RMAN CATALOG command adds metadata about supported existing backup files and copies to the RMAN repository. It is not the command
that creates the recovery-catalog schema, and it is not a synonym for REGISTER DATABASE. Registration enrolls a target database so that
the recovery catalog can maintain its metadata. CATALOG examines an existing file and records the information RMAN can derive from it.
Examples of objects that can be cataloged include a backup piece, data file copy, control-file copy, or archived redo log. RMAN can also search from a specified location and offer matching files for cataloging:
CATALOG BACKUPPIECE '/u03/rman/prod_bkp_01.bkp';
CATALOG START WITH '/u03/rman/' NOPROMPT;
Cataloging is useful after backup files are moved, when user-managed copies must be introduced to RMAN, or when catalog information must be reconstructed from surviving storage. It records metadata, but it does not establish that every block is readable or that an operating-system copy was made while the source file was in a valid state. Those are separate validation and recovery-readiness questions.
The preferred response to catalog database loss is to recover that database from a tested backup. Reconstructing a catalog is a fallback when normal recovery is impossible, not a routine substitute for protecting it. A logical Data Pump export of the catalog schema can supplement physical protection, but it does not replace a complete database recovery strategy.
If catalog recovery is impossible, an administrator can create a new catalog, register the target databases, resynchronize metadata still present in their control files, and catalog backup files that remain discoverable. This process has limits. Metadata that aged out of the target control files may be unavailable, and catalog-only objects such as stored scripts cannot necessarily be reconstructed from backup media.
Oracle 26ai creates a base recovery catalog with the RMAN CREATE CATALOG command after the schema owner and required privileges have
been prepared. Historical procedures that invoke catrman.sql or rely on generic CONNECT and RESOURCE grants are
not appropriate for the current lessons.
RMAN offers related commands that answer different operational questions. Choosing the correct command makes the output easier to interpret and reduces the temptation to treat every repository listing as evidence of recoverability.
| Command | Primary question | Typical examples |
|---|---|---|
LIST |
Which backup objects, archived logs, incarnations, or scripts are recorded? | LIST BACKUP SUMMARY, LIST INCARNATION |
REPORT |
What does RMAN infer about schema history, backup need, obsolescence, or unrecoverable changes? | REPORT SCHEMA, REPORT OBSOLETE |
SHOW |
Which persistent RMAN configuration settings are in effect? | SHOW ALL, SHOW RETENTION POLICY |
LIST BACKUP SUMMARY;
REPORT SCHEMA;
REPORT NEED BACKUP;
REPORT OBSOLETE;
SHOW ALL;
REPORT OBSOLETE evaluates backup records against the configured retention policy. It does not delete the corresponding files.
Likewise, LIST BACKUP displays repository records; it does not prove that every listed piece is available and readable. Repository
maintenance, validation, restore testing, retention, and physical deletion are related controls, but they are not interchangeable operations.
A stored script is a named sequence of RMAN commands maintained in the recovery catalog. A local stored script is associated with the current registered target database. A global stored script is available for use with any target database registered in the catalog. A command file, by contrast, is stored on a file system and is available only where an RMAN client can access that file.
The following statement illustrates the definition of a global script. The later lessons will examine creation, replacement, display, deletion, and execution in detail:
CREATE GLOBAL SCRIPT backup_database
{
BACKUP DATABASE PLUS ARCHIVELOG;
}
The RUN command groups RMAN job commands within braces. When RMAN reads the closing brace, it compiles the command list into job steps
and begins execution. A stored script is executed with EXECUTE SCRIPT inside a RUN block:
RUN
{
EXECUTE SCRIPT backup_database;
}
Stored scripts support consistency, but they do not provide scheduling, version control, approval, credential protection, job monitoring, log retention, or restore assurance by themselves. Production automation needs those controls around the RMAN script. Passwords should not appear in command lines, stored scripts, screenshots, or logs.
For routine human-readable output, LIST, REPORT, and SHOW are usually the most direct tools. SQL queries become
useful when an administrator needs filtering, aggregation, integration with monitoring, or reporting across several registered databases. Oracle
documents recovery catalog views whose names generally begin with RC_, such as RC_BACKUP_PIECE.
An RC_ view can contain metadata for multiple registered databases and their incarnations. The corresponding target-database
V$ view reports the target's own repository information. Catalog-wide queries must therefore identify the intended database and
incarnation with documented identifiers such as DB_KEY, DBINC_KEY, or DBID. Omitting that scope can combine
unrelated rows and produce a report that looks plausible but is operationally wrong.
The recovery catalog views are designed to expose catalog information, but they are not normalized or optimized as a general reporting schema. Use the documented columns and joins, and never update the underlying internal catalog tables directly.
RMAN can send backups to supported cloud storage through a provider-specific system backup to tape (SBT) library. The SBT name describes the interface used by RMAN; it does not mean that cloud backups are written to physical tape.
RMAN command -> configured SBT channel -> provider-specific Oracle module -> cloud storage or recovery service
The historical Oracle Secure Backup Cloud Module name was associated primarily with the Amazon S3 path. Current Oracle 26ai documentation distinguishes the cloud destinations and their modules instead of presenting the old name as a generic multi-cloud product.
| Destination | Current Oracle offering | Integration model |
|---|---|---|
| OCI Object Storage | Oracle Database Cloud Backup Module for OCI | RMAN SBT channel using the OCI module |
| Azure Blob Storage | Oracle Database Cloud Backup Module for Azure | RMAN SBT channel using the Azure module |
| Amazon S3 | Oracle Database Cloud Backup for Amazon S3 | RMAN SBT channel using the Amazon S3 offering |
| OCI Autonomous Recovery Service | Oracle Database Zero Data Loss Cloud Protect for Autonomous Recovery Service | Oracle-managed recovery-service workflow |
The selected module must be installed or configured according to its provider-specific documentation. Authentication files, wallets, channel parameters, installer packages, supported database release updates, and licensing differ by destination and can change over time. A course example for OCI must not be silently reused for Azure or Amazon S3.
A cloud destination does not automatically guarantee lower cost, high availability, regulatory compliance, or successful disaster recovery. Design decisions must account for bandwidth, latency, storage class, retention, immutability, regional placement, retrieval time, egress charges, encryption, key management, and access controls. Many environments retain fast local recovery copies or disk staging as part of a hybrid strategy. Each design must be validated through monitored backup jobs and tested restores.
Oracle 26ai continues to support the recovery catalog and stored RMAN scripts while improving catalog connection management. When RMAN uses a recovery catalog, it automatically disconnects from the catalog database before performing a target database backup. Releasing an unused catalog connection reduces load on the catalog database server.
Catalog upgrade workflows also provide maintenance-mode facilities for identifying blocking or waiting catalog connections and terminating them when the documented procedure requires it. These controls matter in shared catalog environments, but they do not change the central operating principle: protect the catalog, synchronize metadata deliberately, and verify backup storage independently.
The older Oracle Database 11g Release 2 feature list does not define this module. Features such as database duplication, tablespace point-in-time recovery, block repair, platform conversion, and advanced compression belong in lessons that teach those operations. Here, the relevant 26ai changes are current catalog behavior, current command syntax, and current provider-specific cloud integration.
In the next lesson, you will use the RMAN CATALOG command to inspect supported existing backup files and add their metadata to the
RMAN repository.