Proving they work.
WHY IT MATTERS MOST
An untested backup is an assumption, and the assumption fails at the worst moment.
WHAT COMMONLY GOES WRONG
Backups that stopped running months ago Backups excluding something essential Backups that cannot be decrypted Restores that take far longer than anyone expected Media that has degraded
WHAT TO TEST
A full restore, to separate infrastructure, on a schedule.
WHAT TO VERIFY
That the restored system works, not merely that files appeared.
WHAT TO MEASURE
How long the restore actually takes, end to end.
WHY THAT FIGURE MATTERS
It is your real recovery time, regardless of what the plan claims.
WHAT TO TEST PERIODICALLY
Restoring a single file, which is the commonest real request Restoring a database to a point in time Restoring an entire system
WHAT TO DOCUMENT
The restore procedure, in enough detail to follow under pressure.
WHAT TO RECORD
Each test, its date, and its result.
WHAT TO REVIEW
Whether the backup still covers everything, after infrastructure changes.
WHY
New systems are routinely added without being added to backup.
