Why green backup jobs are not enough
Backup tools report whether a job finished. They do not tell you whether the archive is complete, whether the database dump is consistent, or whether anyone remembers the decryption key. Those gaps appear only when you try to restore.
A monthly drill that fits a small team
Pick one system each month and rotate through them. Restore the latest backup to a separate machine or container, never over production.
- Restore the most recent backup to an isolated host.
- Start the application or database and run a basic health check.
- Compare record counts or file counts against production.
- Time the restore from start to usable system.
- Write down every manual step that was not already documented.
Keep backups in more than one place
Keep at least one copy off the server it protects and outside the same provider account. Encrypt it, and store the key somewhere the on-call person can reach without the failed system.
What to record
A short log per drill is enough: date, system, backup used, restore time, problems found and fixes made. Over a year this log becomes the evidence that recovery works, and it shows which systems need attention.
Takeaway
Schedule the drill, rotate the systems, and fix what the drill exposes. Recovery is a practised skill, not a setting.
Frequently asked questions
How often should backups be tested?
Test at least one system every month and every critical system at least once a quarter, by restoring it to an isolated host.
What is a restore drill?
A planned exercise where a team restores a backup to a separate environment, checks it works and records how long it took.
Where should backups be stored?
Keep at least one encrypted copy off the protected server and outside the same provider account.