Home  ›  Glossary  ›  Instant Recovery

What is instant recovery?

Instant recovery is a restore method that mounts a backup copy of a virtual machine disk or database directly from backup storage and starts the workload from it. Reads come from the backup, while new writes go to a separate temporary area. The workload is usable in minutes, before a full copy back to production finishes.

How does instant recovery work on the storage side?

  1. Mount: the recovery software presents the restore point to the hypervisor or database server as a disk or datastore, without copying it first.
  2. Read on demand: the operating system and applications request blocks as they need them, and the backup storage serves those blocks directly.
  3. Redirect writes: changes are written to a separate delta area, often on faster storage, so the restore point itself stays unchanged and reusable.
  4. Migrate: in the background, the data moves to production storage. When the move completes, the delta is merged and the backup storage leaves the data path.

A full restore copies every block before the workload starts. Instant recovery reverses the order: the workload starts first and the copy finishes while it is in use.

Why does backup storage read speed decide the result?

Backup storage is usually built for large sequential writes. A running workload asks for many small, scattered reads, and each slow read is visible to users as a stalled boot or a timed-out application. The figures that matter are random-read performance and latency under load, not the sequential throughput on a datasheet. Slow disks, or storage without enough performance for small reads, make workloads sluggish even when the recovery software works as designed.

How many workloads can run at once?

One mounted machine proves the mechanism. A ransomware recovery may involve dozens or hundreds, all reading from the same storage while it also carries migration traffic. Latency rises with each additional workload until all of them slow down. That limit is a measurable property of the backup storage, found by mounting increasing numbers of workloads and watching boot and response times. It sits in the recovery plan next to restore throughput for full restores.

Which restore points are safe to mount?

Starting a workload from a restore point puts that point into service. A point taken after an attacker arrived brings the attacker's tools with it. Copies held under an immutable backup lock cannot have been altered after they were written, and mounting first into an isolated network, as in clean room recovery, checks whether the content itself is clean.

What else is instant recovery used for?

It lowers the cost of restore testing. Mounting a restore point in a sandbox takes minutes and no production capacity, so more points can be checked more often than full restores allow.

Frequently asked questions

Does instant recovery change the backup?

No. Writes go to a separate delta area, so the restore point stays as it was and can be reused or discarded along with the temporary data.

Is instant recovery the same as failover?

No. Failover switches to a continuously updated replica. Instant recovery starts from an earlier backup, so it loses changes made since that backup, but needs no standby infrastructure.

Does it work for databases and file shares as well?

Yes. Many products can mount a database or file share directly from the restore point, and the same read-performance limits apply.