The automation that matters most.
WHAT TO AUTOMATE
Taking the backup Verifying it completed Transferring it off the system Removing old copies according to retention Alerting on failure
WHAT NOT TO AUTOMATE BLINDLY
The restore.
WHY
Restoring over live data by accident is catastrophic.
WHAT TO AUTOMATE ABOUT RESTORES INSTEAD
A restore into a separate environment, run regularly to verify the backup works.
WHY THAT MATTERS
An untested backup is an assumption.
WHAT TO VERIFY IN A RESTORE TEST
That the data is complete That the application runs against it How long it took
WHAT TO RECORD
The duration, which is your actual recovery time.
WHAT TO ENCRYPT
Backups, which contain everything.
WHERE TO SEND THEM
Somewhere the source system cannot reach or delete.
WHAT TO ALERT ON
A backup that did not run A backup smaller than expected
WHY THAT SECOND CHECK
A backup that completes but captures nothing is the worst kind.
