Virtual machine backup glossary

Short definitions for the words these guides use. Each term also appears in the visible list below.

Virtual machine backup glossary

Use the definitions as a map, then read the guide that applies to your hypervisor. The same word can mean a checkpoint in one product and a backup job in another.

If a vendor screen uses a different label, match it to the failure domain: does the copy survive the loss of the original disks?

Snapshot
A differencing point stored with the original virtual disks. It is not an independent backup.
Backup
A separate copy of a virtual machine that can be registered or imported without the original datastore.
RPO
Recovery point objective: how much recent work you can afford to lose.
RTO
Recovery time objective: how long you can wait before the guest must serve traffic again.
Quiesce
A pause that lets applications flush writes before a snapshot or backup.
Crash-consistent
A copy of disks as they were, without an application flush. The guest may need file-system repair.
Application-consistent
A copy taken after the guest flushed databases and logs.
Off-site copy
A backup kept in another failure domain, such as another site or a separate account.

Questions

Where should the copy live?

On storage that does not share the production datastore, rack power, or backup admin password.

What proves the job worked?

A boot of the restored guest on an isolated network, plus a recent application record.

How long do I keep points?

Daily for two to four weeks, weekly for a few months, and monthly for the period your contract names.