Veeam Explorer for Oracle and Oracle RMAN Integration in v13
Veeam Explorer for Oracle and Oracle RMAN Integration in v13
1. Two Backup Methods, One Recovery Console
Veeam provides two distinct methods for backing up Oracle databases. The first is image-level backup with application-aware processing, where VBR takes a VM-level or agent-level backup and handles Oracle archived redo logs as a guest processing subtask. The second is the Veeam Plug-in for Oracle RMAN, which integrates directly with Oracle's native Recovery Manager and sends backup data to a Veeam repository through the RMAN SBT (System Backup to Tape) interface.
Both methods feed into the same recovery tool: Veeam Explorer for Oracle. Explorer can restore databases from either backup type through a single interface. The key difference is what each method captures and what recovery options each enables.
| Capability | Image-Level Backup | RMAN Plug-in |
|---|---|---|
| Backup scope | Entire VM or volume (includes OS, Oracle binaries, database) | Database only (data files, control files, archived logs) |
| Point-in-time restore | Yes (with archived log processing enabled) | Yes (with archived log backup enabled) |
| Individual PDB restore | No (entire CDB only through Explorer) | Yes (via native RMAN commands against Veeam repository) |
| Supported platforms | VMware, Hyper-V, AHV (VM), Windows/Linux (agent) | Windows, Linux, Solaris, AIX |
| RAC support | VM-level only | Full RAC support (install plug-in on all nodes) |
| Encrypted database support | No (TDE not supported by Explorer) | Yes (TDE supported through RMAN) |
| Instant recovery | Yes | Yes (from Explorer) |
| Oracle home backup | Included in image | Not included (back up Oracle home separately with Agent or image backup) |
You can run both methods against the same Oracle server. If you do, be careful with archived log handling. Only one tool should manage log deletion. If both the image-level job and the RMAN plug-in are configured to delete archived logs, you will break the log chain for one or both backup methods.
2. Image-Level Backup with Archived Log Processing
When you back up an Oracle VM with application-aware processing enabled, VBR puts the database into backup mode before taking the VM snapshot, ensuring a consistent capture of the data files. After the snapshot, VBR exits backup mode. This creates a transactionally consistent restore point.
For point-in-time recovery, you need archived redo log backups. These are configured in the backup job's Guest Processing settings under the Oracle tab, the same pattern as SQL Server transaction log backup.
Configuring Archived Log Backup
- 1Edit the backup job. Navigate to Guest Processing.
- 2Enable Application-aware processing.
- 3Click the Oracle VM in the list and select Edit. Switch to the Oracle tab.
- 4Select Backup logs periodically. Set the interval (for example, every 15 minutes).
- 5Configure log retention and whether to delete archived logs from the Oracle server after backup.
The database must be running in ARCHIVELOG mode. If it is in NOARCHIVELOG mode, archived redo logs are not generated and point-in-time recovery is not possible. Check the mode:
If it says NOARCHIVELOG, switch it. This requires a controlled shutdown and restart:
Data Guard and Archived Log Deletion
If the Oracle database participates in a Data Guard configuration, do not set aggressive log deletion intervals. Data Guard needs time to transport logs to the standby before they are purged. Setting "Delete logs older than" to less than 24 hours is not recommended in Data Guard environments.
3. Veeam Plug-in for Oracle RMAN
The RMAN plug-in works differently from image-level backup. It installs a shared library (libOracleRMANPlugin.so on Linux, OracleRMANPlugin.dll on Windows) that RMAN loads as an SBT media management device. When RMAN runs a backup, data flows through the plug-in directly to a Veeam backup repository. No VM snapshot. No image-level capture. Just the Oracle data files, control files, SPFILEs, and archived logs.
This is the right choice for physical Oracle servers, Solaris and AIX deployments (which Veeam Explorer does not support for restore), RAC clusters that need plug-in deployment on every node, and environments where TDE (Transparent Data Encryption) is in use.
RMAN commands look normal. The only difference is the channel allocation, which points at the Veeam SBT library instead of disk or tape:
The plug-in handles communication with the Veeam backup server and repository. RMAN does not need to know anything about Veeam infrastructure. It just writes to the SBT channel and the plug-in routes the data.
4. RMAN Plug-in: Managed Mode vs Standalone Mode
The plug-in can operate in two modes. Managed mode means the VBR server deploys, configures, and updates the plug-in automatically through protection groups. Standalone mode means you install and configure the plug-in manually on the Oracle server and it registers itself with VBR.
| Aspect | Managed Mode | Standalone Mode |
|---|---|---|
| Plug-in deployment | Automatic via protection group | Manual install on Oracle server |
| Configuration | Backup policy created in VBR console | RMAN commands executed on Oracle server |
| Auto-update | Yes (configurable) | Manual |
| Backup scheduling | VBR policy-driven | Oracle scheduler, cron, or manual |
| Best for | Centralized management, many Oracle servers | DBA-controlled environments, custom RMAN scripts |
In managed mode, you create a protection group for your Oracle servers, select "Install application plug-in" with the Oracle RMAN option, then create a backup policy that defines schedule, retention, repository, and archived log handling. The VBR console drives everything.
In standalone mode, the DBA retains full control over RMAN scripts and scheduling. VBR sees the backup jobs that the plug-in creates but cannot modify them. This mode is preferred in environments where the Oracle team has established RMAN workflows and does not want the backup team controlling database-level operations.
5. Veeam Explorer for Oracle: Recovery Operations
Explorer for Oracle supports the same core operations as Explorer for SQL Server: restore to latest state, point-in-time restore with archived log replay, instant recovery, and database publishing. It works against both image-level and RMAN plug-in backups.
The recovery process varies slightly between Windows and Linux Oracle servers. On Windows, Explorer deploys a non-persistent runtime component called Veeam Oracle Restore Service to the target server. On Linux, Explorer connects via SSH and runs operations directly. In both cases, VM disks from backup are mounted via iSCSI (Windows) or FUSE and loop device (Linux).
6. Point-in-Time Restore with Archived Log Replay
Point-in-time restore for Oracle works on the same principle as SQL Server: select a timestamp, and Explorer replays archived redo logs from the closest prior restore point to reach that timestamp.
- 1Open Veeam Explorer for Oracle from the VBR console: Backups > right-click restore point > Restore application items > Oracle databases.
- 2Browse the database hierarchy. Select the target database.
- 3Right-click and select Restore database > Restore to a point in time.
- 4Select the target timestamp within the available recovery window.
- 5Choose restore target: original server or alternate server. Provide OS and Oracle credentials.
- 6Explorer copies database files and archived redo logs from the backup repository to the target server, replays the logs to the specified point, and starts the recovered database.
On Windows targets, temporary files (scripts, archive logs, cache) are stored in C:\Windows\Temp on the target and write cache is on the mount server. On Linux targets, temporary files go to /var/tmp or /tmp, whichever has more free space.
Backup Copy Jobs and Cloud Repositories
Point-in-time restore is not supported for backups stored in the DR site by backup copy jobs or for backups stored in cloud repositories. These repositories cannot serve as a destination for archived log files during restore. You can restore to the latest restore point state only.
7. Instant Database Recovery
Instant recovery for Oracle follows the same architecture as SQL Server instant recovery. Explorer sends a command to the Veeam Mount Service, which starts the database directly from backup via mount. The database comes online in minutes. A background copy migrates data to the target server's native storage while the published database serves reads and writes.
The mount location is C:\VeeamFLR on Windows and /run/media on Linux. Write cache location differs: Windows uses the mount server (C:\ProgramData\Veeam\Backup\IRCache\ by default), Linux uses /var/tmp on the target server.
During switchover, the recovery service stops the published database, synchronizes remaining differences using redo logs, drops the published database, and starts the recovered database. The session survives network disruptions with 10 automatic retries at 5-minute intervals.
If the source database uses ASM (Automatic Storage Management), the target server must also have an ASM group configured for instant recovery to work.
8. Database Publishing
Publishing attaches an Oracle database from backup to the target server as a temporary live mount. No background copy. No switchover. The database is available for queries and data extraction as long as the Explorer session is open.
This is the recommended workaround for individual PDB restore. Veeam Explorer cannot restore individual pluggable databases from a CDB backup. But you can publish the entire CDB from backup, then use native RMAN commands against the published database to back up and restore specific PDBs to your production environment.
Data Guard configurations cannot be published as a complete configuration. Individual databases from a Data Guard setup can be published as standalone Oracle databases, but the Data Guard infrastructure (primary/standby relationships, log transport) is not preserved.
9. Restoring from RMAN Plug-in Backups
RMAN plug-in backups appear in the VBR console under Backups > Disk (Application Plug-ins). You can launch Explorer against these backups exactly as you would against image-level backups. Explorer opens, mounts the backup, and presents the database hierarchy.
For advanced restore scenarios that Explorer does not support (individual PDB restore, restore with different database name and settings, cross-platform restore to Solaris/AIX), you use native RMAN commands directly on the Oracle server with the Veeam SBT channel. The plug-in makes Veeam's repository appear as an RMAN media device. RMAN handles the rest using its native restore and recover commands.
To get the correct FORMAT string for your environment, check the Veeam plug-in logs on the Oracle server at /tmp/veeam_plugin_logs/oracle/. The task log contains the exact channel allocation and format string used during backup.
Restoring to Another Server
When restoring RMAN plug-in backups to a different Oracle server, the Veeam Plug-in for Oracle RMAN must be installed on the target server and configured to communicate with the same VBR server and backup repository. The account connecting the target plug-in to VBR needs either Backup Administrator or both Backup Operator and Restore Operator roles.
10. CDB, PDB, and Data Guard Considerations
Multitenant Container Databases (CDB). Veeam Explorer can restore entire CDBs. It cannot restore individual PDBs through the Explorer interface. This is a known limitation that is on Veeam's roadmap. The current workaround is to publish the CDB and use native RMAN commands to restore specific PDBs, or to use the RMAN plug-in with native RMAN restore commands directly.
Refreshable PDBs. Explorer does not support restore, publishing, instant recovery, or export of refreshable pluggable databases. Use the RMAN plug-in for these.
Data Guard. Explorer can restore individual databases from a Data Guard configuration. It cannot restore or publish the complete Data Guard configuration (primary plus standby plus log transport services). Restored databases come up as standalone instances with no Data Guard relationships.
RAC. For RMAN plug-in backups, restoring to the original RAC configuration requires restoring with original name and settings. Otherwise the database is restored as a standalone instance. Install the RMAN plug-in on all RAC nodes for full coverage.
11. Limitations and Gotchas
Solaris and AIX are not supported by Explorer. Oracle databases on Solaris or IBM AIX cannot be restored through Veeam Explorer. Use the RMAN plug-in with native RMAN restore commands instead.
Encrypted databases (TDE) are not supported by Explorer. If Transparent Data Encryption is enabled, Explorer cannot restore the database. Use the RMAN plug-in, which supports TDE through RMAN's native encryption handling.
ARCHIVELOG mode is mandatory for point-in-time restore. If the database is running in NOARCHIVELOG mode, you can only restore to the exact state of a restore point. No log replay. No granularity between restore points.
After restore, auto-start is disabled. Explorer disables Oracle auto-start after a database restore. On Windows, you need to set the ORA_SID_AUTOSTART registry key back to TRUE. On Linux, edit the oratab file to re-enable auto-start for the restored instance.
Btrfs and ZFS mount limitations. Instant recovery, data restore, and data publishing to the original server will fail if the Oracle data files are on Btrfs or ZFS volumes. The issue is that two Btrfs or ZFS volumes with identical IDs cannot be mounted on the same machine simultaneously (one from backup, one native).
Linux SSH shell requirement. The account used to connect to Linux Oracle servers for restore must have a valid shell (bash). If the Oracle service account has /sbin/nologin or /bin/false as its shell, Explorer will fail to connect.
RMAN compression and Veeam compression. Do not enable both simultaneously. If you use RMAN compression during backup, disable Veeam Data Mover compression for the RMAN plug-in job. Double compression wastes CPU and can degrade backup performance.
Temp Tablespace sizing. When running RMAN jobs through the Veeam plug-in, Oracle may use Temp Tablespace resources for statistical queries. Ensure the Temp Tablespace is adequately sized to prevent "insufficient temporary tablespace" errors during backup operations.
$ORACLE_HOME/bin must be in PATH. The RMAN plug-in needs to find oci.dll (Windows) or the equivalent shared library in the Oracle server process PATH. If it is missing, you get ORA-19554 and ORA-27002 errors about media management library failures. Add $ORACLE_HOME/bin to the PATH environment variable for the Oracle server process.
Key Takeaways
- Two backup methods: image-level with archived log processing (VM/agent) and RMAN plug-in (native RMAN integration). Both restore through Veeam Explorer for Oracle.
- The RMAN plug-in uses Oracle's SBT interface to write directly to Veeam repositories. RMAN commands are standard except for channel allocation.
- Managed mode lets VBR deploy and control the plug-in. Standalone mode gives DBAs full RMAN script control.
- Explorer cannot restore individual PDBs. Workaround: publish the CDB, then use native RMAN commands to restore specific PDBs.
- TDE-encrypted databases and Solaris/AIX platforms are only supported through the RMAN plug-in, not Explorer.
- Instant recovery works on both Windows and Linux Oracle servers with OS-specific mount paths and write cache locations.
- Data Guard databases restore as standalone instances. The Data Guard configuration itself is not preserved.
- Only one tool should manage archived log deletion. Running both image-level and RMAN log management will break the log chain.