An untested backup is an assumption rather than a guarantee. And assumptions share one property: you discover whether they hold at the worst possible moment — the day you have to restore.
Short answer: what is a restore drill?
A restore drill means restoring your existing backup into a separate environment, without touching production, to measure two things: how long recovery takes and whether the recovered data is complete. The fact that backups are running answers neither question; it only shows that files exist somewhere. For most businesses a drill takes a few hours, and run twice a year it reduces the disaster scenario to a known duration.
The five steps of a drill
| Step | What is done | What is measured |
|---|---|---|
| 1. Choose a scenario | A clear assumption such as "the server vanished last night" | Clarity of scope |
| 2. Prepare the environment | A separate test environment; production untouched | Preparation time |
| 3. Restore | Files, database and configuration brought back | Total duration |
| 4. Verify | Does the site open, is data complete, do flows work | Integrity |
| 5. Record | Duration, gaps and obstacles written down | Input for the next drill |
The most commonly skipped check sits at step four: the site opening is not enough. Login, form submission, the payment step and outbound e-mail must each be tried. An installation whose files came back but whose database connection still points at the old server does not count as restored.
Two numbers to measure
- Recovery time. From noticing the incident to the system working again. This figure determines your downtime cost directly and forms the basis of your continuity plan.
- Data loss window. The gap between the last backup and the moment of the incident. Backing up daily means losing up to a day's data in the worst case. Whether that window is acceptable is a business decision, not a technical one.
Once both are written down, the backup discussion leaves the vague "is it enough" and becomes a concrete question of acceptability. We covered both concepts in our backup and disaster recovery article.
Typical findings from a drill
- Incomplete scope. Files present, database absent; or database present, uploads folder missing. The most frequent finding.
- Configuration not carried over. Server settings, certificates, cron definitions and mail configuration fall outside the backup. The site returns but does not work.
- The backup is corrupt or partially written. The archive will not open, or some tables are empty. Only attempting to open it reveals this.
- Missing credentials. Whoever must restore has no access to the provider panel; the delay is administrative rather than technical.
- Recovery takes far longer than expected. Downloading large files and importing a database can take multiples of the estimate.
- The encryption key cannot be found. The backup is encrypted and the key was stored only on the old system.
Learned during a real incident, each of these is costly; learned in a drill, each is merely a correction item.
How often?
- At least twice a year for a full drill (files + database + configuration).
- Quarterly light verification: backup exists, its size is right, and it opens.
- After every major change: server migration, version upgrade, new system added.
- Whenever the backup configuration changes, a verification is mandatory.
Netişlem expert view: the most common finding is incomplete scope
In the drills we run, the most frequent problem is not a corrupt backup but an incomplete one. Files are backed up regularly while the database lives in another system nobody has checked. Or both exist, but mail accounts and server configuration sit outside scope. In that state, "we have backups" is technically true and practically misleading.
Our second observation is that recovery time is almost never measured. Businesses say "we would restore from backup" without knowing how many hours that takes. When measured, it usually comes out markedly above expectation — and that changes the continuity plan directly.
Third, drills are treated as pass-or-fail tests. Finding problems is the point; a drill that surfaces nothing means either you prepared exceptionally well or you did not look deeply enough. Drills producing no findings usually belong to the second group.
Drill checklist
- Scenario written with clear scope.
- Test environment fully separate from production.
- Restore performed only from the backup (no copying from production).
- Files, database and configuration restored together.
- Site opens; pages beyond the home page checked too.
- Login, forms and payment flow tried.
- Outbound e-mail tested.
- Scheduled task definitions checked.
- Total duration recorded.
- Data loss window calculated.
- Gaps and obstacles written down with owners and dates assigned.
- Next drill scheduled.
Frequently asked questions
Does a drill put production at risk?
Not when done properly. Restoration happens in a separate environment and production is never touched. One rule removes the risk: no changes are made to the live system during the drill.
My provider takes backups — do I still need a drill?
Yes. A provider holding backups does not mean you can restore; scope, timing and request processes may differ. A drill can use the provider's backup — what you measure is how long recovery takes with it.
Is this necessary for a small site?
The smaller the scale, the shorter the drill; on a small site it may take an hour. But the necessity does not change: whether a backup works is a question independent of site size.
How many people should be able to restore?
At least two. Recovery capability resting on one person is void when that person is unreachable. Required access and steps must be written, and the second person should take part in the drill.
What if the drill finds problems?
Findings are the drill's success. Assign an owner and date to each, and run a short verification after correction. What matters is that findings get closed rather than staying on a list.
Conclusion
Backup is measured not by the number of copies but by the ability to return. A few hours twice a year turns "we have backups" into two concrete figures: how many hours to recover, and how much data is lost. A business that knows both does not have to decide during a crisis.
If you would like help planning your drill, get in touch. You can review our corporate backup, SLA guarantee and server management solutions.