Backups nobody has restored
Almost every estate has backups. Far fewer have a restore that has been performed this year, and fewer still have one performed by someone other than the person who built it. Retention is usually whatever the default was. The recovery time objective exists as a number in a document and has never been measured.
What the work involves
- Establish what must survive what: which data, which systems, and how much loss and downtime the business can actually absorb.
- Backups verified by restoring them — into an isolated environment, on a schedule, with the result recorded.
- Retention matched to the obligation you are actually under rather than to a default, and priced, because retention is a recurring cost. a rehearsed rollback, and retention extended to one year →
- Recovery runbooks written to be executed by someone who did not write them, and rehearsed at least once.
- Immutability and separation where the threat model calls for it: a backup reachable with the same credentials as production is not a backup from a ransomware perspective.
How the work goes
Objectives
Recovery point and recovery time agreed per system with the people who own the consequences, not assumed by engineering.
Verify
Restore what exists today and time it. This is where the difference between believed and actual recovery time shows up.
Close the gaps
Fix what the rehearsal exposed, then schedule the rehearsal so it keeps happening after we are gone.
Is this the right call?
A good fit when
- Teams who have backups and no evidence they work
- Anyone facing an audit, an insurer or a customer questionnaire that asks about DR
Not us when
- Producing a DR document to satisfy a checkbox without testing anything — we will decline that one