Recovery Plan Fails After Region Outage
Contoso runs a three-tier app in East US. VMs are backed up with Azure Backup to a Recovery Services vault configured with Geo-Redundant Storage. The vault's soft delete and enhanced security features are enabled. After a regional outage, an engineer attempts a cross-region restore of a 512 GB VM. The restore fails with 'The backup data is not available in the secondary region.' The vault shows the last successful backup 40 minutes before the outage. The RPO target is 15 minutes and the RTO target is 2 hours. The team also reports that a separate SQL database on a VM cannot be restored to a point in time older than 7 days, although the retention policy states 35 days.
1. Diagnose why the cross-region restore fails and what must change to meet the RPO and RTO targets.
Read the worked answer
The key fact is that geo-redundant storage alone does not make backup data available in the paired region for restore. Cross-region restore must be explicitly enabled on the vault, and even when enabled, only backups taken after the feature was enabled are replicated to the secondary region. The 40-minute-old backup predates enablement or was never replicated, so the restore fails.
Reasoning chain:
- Storage redundancy (GRS) protects the storage account, not the restore capability. The vault setting Cross Region Restore controls availability of recovery points in the secondary region.
- Enable Cross Region Restore, then wait for new recovery points to replicate. Old points will not become available retroactively.
- The RPO of 15 minutes is not met by a 40-minute gap; increase backup frequency for the VM (for example, multiple backups per day via an enhanced policy) so the interval is under 15 minutes.
- RTO of 2 hours requires pre-staged capacity in the secondary region; a 512 GB restore can exceed 2 hours if disks must be created cold. Validate with a test restore.
- For the SQL point-in-time issue, 7 days suggests only the default short-term recovery points exist; the 35-day retention needs a long-term retention policy or the SQL backup configuration must be corrected.
2. What verification steps would you perform before declaring the continuity plan fixed?
Read the worked answer
Verification should be evidence-based, not assumption-based:
- Confirm Cross Region Restore shows Enabled on the vault and that the secondary region matches the intended pair.
- List recovery points in the secondary region and confirm their timestamps are within the RPO window.
- Perform a test restore of the 512 GB VM into an isolated resource group and measure elapsed time against the 2-hour RTO.
- For SQL, query the actual recovery point range rather than trusting the policy display, and confirm long-term retention recovery points exist for days 8 through 35.
- Run a documented failover drill and record results; a plan that has never been exercised is not verified.