With Instant VM Recovery, you can immediately restore a VM into your production environment by running it directly from the backup file. Instant VM recovery helps improve recovery time objectives (RTO), minimize disruption and downtime of production VMs. It is like having a "temporary spare" for a VM: users remain productive while you can troubleshoot an issue with the failed VM.
When Instant VM Recovery is performed, Veeam Backup & Replication uses the Veeam vPower technology to mount a VM image to an ESXi host directly from a compressed and deduplicated backup file. Since there is no need to extract the VM from the backup file and copy it to production storage, you can restore a VM from any restore point (incremental or full) in a matter of minutes.
The archived image of the VM remains in read-only state to avoid unexpected modifications. By default, all changes to virtual disks that take place while the VM is running are logged to auxiliary redo log files residing on the NFS server (backup server or backup repository). These changes are discarded as soon as a restored VM is removed, or merged with the original VM data when VM recovery is finalized.
To improve I/O performance for a restored VM, you can redirect VM changes to a specific datastore. In this case, instead of using redo logs, Veeam Backup & Replication will trigger a snapshot and put it to the Veeam IR directory on the selected datastore, together with metadata files holding changes to the VM image. Redirecting VM changes improves recovery performance but makes Storage vMotion not possible for ESX 4.x and earlier. As a result, you will not be able to use Storage vMotion to finalize Instant VM Recovery.
To finalize Instant VM Recovery, you can do one of the following:
- Use Storage vMotion to quickly migrate the restored VM to the production storage without any downtime. In this case, original VM data will be pulled from the NFS datastore to the production storage and consolidated with VM changes while the VM is still running. Storage vMotion, however, can only be used if you select to keep VM changes on the NFS datastore without redirecting them. Note that Storage vMotion is only available with select VMware licenses.
- Use replication or VM copy functionality of Veeam Backup & Replication. In this case, you can create a copy of a VM and fail over to it during the next maintenance window. In contrast to Storage vMotion, this approach requires you to schedule some downtime while you clone or replicate the VM, power it off and then power the cloned copy or replica on.
- Use Quick Migration. In this case, Veeam Backup & Replication will perform a two-stage migration procedure — instead of pulling data from the vPower NFS datastore, it will restore the VM from the backup file on the production server, then move all changes and consolidate them with the VM data. For details, see Quick Migration.
In many respects, Instant VM Recovery gives results similar to failover of a VM replica. Both features can be used for tier-1 applications with little tolerance for business interruption and downtime. However, when you perform replica failover, you do not have dependencies on the backup server. And, unlike Instant VM Recovery that provides only limited I/O throughput, replication guarantees full I/O performance.
Beside disaster recovery matters, Instant VM Recovery can also be used for testing purposes. Instead of extracting VM images to production storage to perform regular DR testing, you can run a VM directly from the backup file, boot it and make sure the VM guest OS and applications are functioning properly.