Restoring from Azure Backup

Each successful backup run creates a recovery point — a point-in-time snapshot of the protected resource. Azure Backup retains recovery points according to the backup policy's retention rules. When you need to restore, you select a recovery point and choose a restore type. The available types depend on the resource: for VMs you can do a full VM restore, disk restore, or file-level item recovery. Understanding these options helps you choose the fastest path to recovery.

A full VM restore recreates the entire virtual machine — OS disk, data disks, and network interface — from a recovery point. You can restore to the original location (replacing the existing VM) or to an alternate location (creating a new VM with a different name or in a different resource group or region). Full restore is the slowest option but recovers a completely broken or deleted VM. Cross-region restore is also supported if the vault uses GRS storage.

Disk restore exports the managed disks from a recovery point to a specified storage account without creating a new VM. This is useful when you only need to recover specific data from a disk, or when you want to attach the recovered disk to an existing VM for investigation. After the disks are exported, you can mount them, browse the file system, and copy only the files you need — faster than spinning up a full VM restore.

Item-level recovery (ILR) lets you browse and recover individual files and folders from a VM backup without restoring the entire disk. Azure Backup mounts the recovery point as a local iSCSI target on a Windows or Linux machine. You connect using a provided script, browse the mounted disk as if it were a local drive, and copy only the specific files you need. The mount session expires after 12 hours to limit exposure.

When evaluating backup strategies, two metrics matter: RPO (Recovery Point Objective) is the maximum tolerable data loss expressed as time — if your backup runs daily, your RPO is up to 24 hours. RTO (Recovery Time Objective) is how long the restore process itself takes. Full VM restores can take 30 minutes to several hours depending on disk size. File-level recovery is much faster. Matching your RPO and RTO to business requirements drives your backup schedule and restore strategy choice.

For Azure Files backed up via Azure Backup, you have two restore options. Original location restore overwrites the current share contents with the snapshot. Alternate location restore restores to a different file share or storage account, preserving the current share unchanged. You can also do item-level restore to recover specific files or folders from the snapshot, which is the most common and lowest-impact recovery type for file shares.

Cross-region restore (CRR) allows you to restore VMs to a secondary (paired) region when the primary region experiences an outage. CRR is only available for vaults configured with Geo-Redundant Storage. You must explicitly enable CRR in vault properties — it is not on by default. CRR adds a cost premium but is essential for scenarios where you need to restore workloads in the secondary region before Azure Site Recovery completes failover.

Every restore operation creates a backup job with type Restore that you can monitor in the vault's Backup Jobs blade. Large disk restores can run for hours. If a restore fails partway through (e.g. due to a permissions issue on the target storage account), Azure Backup provides a detailed error message and a link to documentation. You can retry the restore once the underlying issue is fixed — a partially completed restore does not corrupt the recovery point.

When SQL Server is running on an Azure VM and protected with the Azure Backup for SQL Server solution, you get additional restore granularity. You can restore to a specific point in time (based on transaction log backups that run every 15–60 minutes) rather than only to a full backup recovery point. This dramatically lowers RPO for SQL workloads compared to file-system-level VM backup alone.

Many organisations never test their backups until disaster strikes — and then discover the backups are corrupted, incomplete, or take far longer to restore than expected. Best practice requires regular restore drills: restore to a test resource group, verify application functionality, and document the actual RTO achieved. Azure Backup Center's Backup Reports workbook provides a compliance view showing the last successful backup date for every protected item.

Azure Backup for VMs maintains a local instant restore snapshot for 1–5 days (configurable in the policy) in the resource group alongside the VM. Restoring from this local snapshot is much faster than restoring from the vault because data does not need to be transferred from the vault's storage. After the instant restore tier expires, the recovery point is stored only in the vault. Using instant restore for recent accidents significantly reduces your RTO.

Test your understanding of Microsoft Azure Fundamentals (AZ-900) concepts from this lesson.

In this lesson you learned: Azure Backup offers multiple restore types — full VM, disk, and file-level item recovery — each with different speed and granularity trade-offs, cross-region restore requires GRS vaults and must be explicitly enabled, and regular restore drills are essential to verify your actual RTO matches business requirements. Next up we explore Azure Site Recovery for replicating entire workloads to a secondary region.