No sentence starts more recoveries than ‘we had backups’. Not ‘we had no backups’ — the honest doomed, who at least know their position — but the past tense of people discovering, mid-disaster, that their safety net was decorative. After twenty-five years of these conversations, the failures sort into five species. Learn to recognise yours before it matters.
Mirrors, stale jobs, sync traps, single points, untested restores — and the one test that catches them all.
1. The mirror. RAID, a synced NAS, a second drive that copies the first: availability tools mistaken for backups. Mirrors reproduce everything faithfully — deletions, corruption, ransomware — at the speed of light. 2. The stale job. The scheduled backup that silently stopped in March: destination full, credential expired, task disabled by an update — discovered in November. 3. The sync trap. Cloud sync folders that dutifully propagate your local deletion to the ‘backup’ within seconds; version history helps if the plan includes it and you know before it expires. 4. The single point. The backup drive stored beside the computer, sharing every fire, flood, surge and burglary; the NAS that is the only copy. 5. The unrestorable. The backup nobody ever test-restored — encrypted with a lost key, corrupt for months, or missing the one folder that mattered because it was added after the job was configured.
Every species above dies to a single predator: the test restore. Not checking the job ran — restoring an actual file, from the actual backup, to a different machine, and opening it. That one act verifies the copy exists, is current, is readable, is complete, and is reachable without the original system — five verifications in five minutes. Do it quarterly; diarise it; make it someone’s job. The 3-2-1 shape (three copies, two media, one off-site) is the architecture, but the test restore is the inspection — and an untested backup is, statistically speaking, a hope.
Found this page because the backup just failed you? Then two truths, quickly. First: the failed backup and the failed original are now both recovery sources — a stale backup plus a recovered delta from the dead drive frequently reassembles the whole, so preserve both and tell the bench what dates each covers. Second: the machine that discovers a backup failure is usually asked to ‘just try’ things in the following hour — and that hour does more damage than the original disaster. Power down, inventory what copies exist, and let the free diagnostic map the union of what survives.
Not sure which species yours is? The test restore will tell you in five minutes — and the bench can tell you what it costs when it’s already told you the hard way. The diagnostic is free, the quote is fixed in writing, and honest advice on 0131 202 0491 costs nothing at all.