RestoreDrill pulls your real backups into a throwaway sandbox, restores them for real, checks that the data actually came back, then burns the sandbox down. You get a timestamped report your auditor will accept.
Sample drill output. Every line is evidence — and the RTO is measured, not estimated.
Three failures that look identical from the outside — a green checkmark on last night’s job — right up until the day they don’t.
Teams can produce a backup policy. Very few can produce evidence that a restore was ever tested. A policy without test evidence is not an operating control — it is paperwork, and auditors treat it that way.
The job exits 0 every night for eight months. The dump has been truncated since March, because a schema change broke one flag. Nobody notices until the restore that finally matters.
Someone wrote “4 hours” in the DR plan. Nobody has measured it. The real number is whatever it turns out to be, on the worst possible day, while customers wait.
Nothing about how you back up has to change. RestoreDrill only exercises what you already have.
Point the agent at backups you already produce — restic, kopia,
borg, pg_dump, mysqldump, or a plain S3 bucket.
Read-only credentials are enough.
On your schedule, the agent restores a real snapshot into a disposable sandbox on your own infrastructure. Your data never leaves your network, and production is never touched.
Row counts, table counts, file checksums, and a real health check against the restored service — not just “the archive extracted”. Then the sandbox is destroyed.
A timestamped report: what was restored, what was verified, what failed, and the measured RTO. Export it for your auditor, or give them read-only access to the history.
Restore testing is an explicit control. You need evidence on a schedule, not a screenshot from last quarter.
You back up for twenty clients. Proving it per client, every month, is currently a manual afternoon.
No auditor, no compliance deadline — just no appetite for discovering the truth during an outage.
These numbers are a starting hypothesis, not a price list — there is nothing to buy yet. If a tier is priced wrong for the job it does, that is exactly the kind of thing we want to hear on the waitlist form.
Which backup tool gets supported first depends on who signs up. Two questions, then we leave you alone until there is something real to try.