Access Official Technical PDF

Share Summary:LinkedInEmail

Oracle Recovery Appliance cloud backup links an on-premises database protection system with cloud object storage, providing another destination for backup data alongside local disk, tape or a remote appliance. For industrial organizations running business-critical databases, the design is not simply about where backup files reside. It also involves how changes are captured, how data is encrypted and retrieved, and whether network capacity can support recovery objectives.

The architecture described here combines Recovery Appliance with Oracle Cloud services and, in a separate deployment model, Exadata Cloud at Customer. Its core ideas—incremental data movement, continuous redo protection and deliberate recovery planning—are relevant to technical teams evaluating database resilience in hybrid environments.

How Oracle Recovery Appliance cloud backup works

Recovery Appliance uses an RMAN-driven framework to receive database backups. Rather than repeatedly transmitting full backups, protected databases send incremental changes through a process called Delta Push. Real-time redo transport can also send transaction records as they are generated, helping protect changes between scheduled backups.

On the appliance, incoming changes are validated and compressed in a disk-based Delta Store. The system can use those changes to present a virtual full backup for restore operations. This approach separates the logical recovery point from the need to create and move a new physical full backup each time. The appliance can also replicate data to a remote Recovery Appliance or offload backup copies to tape or cloud storage.

In the cloud archive workflow, an incremental backup reaches the on-premises appliance, where a virtual full is created. The appliance obtains an encryption key from an Oracle Key Vault server, encrypts the backup pieces and sends them to cloud object storage. If the storage container uses a tiering policy, the pieces can subsequently move to archive storage.

Plan the restore path, not just the backup path

Cloud retention is useful only if the recovery path is understood and tested. When restoring through Recovery Appliance, the appliance requests the encryption key from Key Vault and retrieves the required backup pieces from cloud storage. If those pieces have been moved to archive storage, they must first be recalled to object storage. The appliance then decrypts the data and sends it to the protected database.

The material also describes a direct restore route in which an eligible database reads backup pieces from cloud object storage. In that workflow, the database must be enrolled as a Key Vault endpoint to obtain the decryption key. The documented process distinguishes object storage from archive storage; direct restore from Oracle Cloud Archive Storage was identified as not then available in the described plan.

These alternatives have practical implications for recovery time objectives (RTOs). Teams should account for network throughput, database size, archive recall, key availability and the chosen restore route when estimating how long recovery may take. A backup destination alone does not establish that a required recovery time can be met.

Operational visibility and deployment considerations

Enterprise Manager reporting was presented as a way to track backup activity at database level, including virtual full, tape and replicated copies. Reports can show destinations, transfer rates, backup types, start and end times, sizes and backup-piece details. Tenant-view monitoring filters Recovery Appliance target metrics by protected database within the Enterprise Manager repository, without adding collections on the appliance.

For deployments involving Exadata Cloud at Customer (ExaCC), the presentation emphasized planning network architecture early. Backup and client networks, available throughput and database size all influence backup and restore behavior. Before deployment, teams should:

  • Engage network and Maximum Availability Architecture specialists during design and testing.
  • Map the network topology and identify possible throughput bottlenecks.
  • Connect network capacity and database size to the required RTO, and flag gaps before production use.
  • Test RMAN network performance and validate the configuration after initial service setup.
  • Review wallet and encryption-key handling, including how it interacts with Transparent Data Encryption and redo shipping.
  • Check the backup API configuration and required database software components against the applicable Recovery Appliance practices.

Deployment experience: ExaCC and Recovery Appliance

Dialog Semiconductor described a configuration in which an ExaCC primary environment sent redo to an on-premises standby system, which in turn shipped redo to Recovery Appliance. This arrangement was intended to avoid placing backup load on the primary. The company reported that its recovery time objective fell from 48 hours to 12 hours after adopting Recovery Appliance. It also reported that a weekly level 0 backup took about 10 hours, compared with 38 hours previously, and that incremental level 1 backup time decreased from more than three hours to around 1.5 hours.

The example illustrates reported results from one deployment, not a guaranteed outcome for other environments. The presentation also described migration using Recovery Appliance backups and Data Guard: a restored database could be added as a standby and switched over, with the reported switchover taking less than five minutes. A roughly 52 TB production database was used in restore and duplication testing. That work surfaced issues involving a network switch port and a restore timeout for a large datafile; both were addressed, underscoring the value of testing realistic database sizes, RMAN settings and network paths before relying on a design.

Make recovery requirements the design starting point

Recovery Appliance and cloud storage can support a layered database protection strategy: incremental backups reduce repeated full-data movement, redo transport captures ongoing transactions, and cloud copies provide an additional storage destination. The benefits depend on the complete operational chain—from network design and key management to archive retrieval, monitoring and restore testing.

For automation and operations leaders, the practical step is to validate recovery scenarios against business requirements rather than treating backup completion as the measure of readiness. Confirm the intended restore route, test it with representative data volumes, and assess whether network and storage behavior can meet the required RTO and recovery point objective. That discipline turns a hybrid backup architecture into a recovery plan teams can execute.

Access Official Technical PDF

About The Author