Blog

A backup isn’t a plan until you’ve restored from it.

Backups fail quietly, and you find out on the worst day. How to rehearse recovery, and the two numbers every business should know.

Published
Reading time
3 min read
Answers
C · What happens when things change?
By
dataSMC

Almost every business has backups. Far fewer have restored from them recently. The difference only matters on one day — the day something is lost — and on that day it is the only thing that matters.

Backups fail in quiet ways. A job that stopped running when a password changed. A drive that filled up months ago. A system added after the backups were set up, and never included. Nothing alerts, because from the backup system’s point of view nothing is wrong. You find out when you need the file.

Backups fail quietly

The dangerous assumption is that a backup job reporting “success” means you can get your data back. It means the job ran. It doesn’t mean the right things were in it, that the copy can be read, or that anyone knows how to put it back.

The only way to know a backup works is to restore from it.

Rehearse the bad day

Recovery is a skill, and skills fade. So rehearse it, on a calendar, before you need it.

Every quarter, restore something small. Pick a file, a mailbox or a folder at random, restore it somewhere safe, and check it opens. It takes minutes and it tests the whole chain: the job ran, the copy is readable, someone knows the steps.

Every year, restore something big. Take one important system and bring it back from its backups, end to end, somewhere it can’t disturb the real one. Time it. Write down every step that surprised you.

The first full rehearsal is always uncomfortable. That is the point: you would rather be uncomfortable on a Tuesday afternoon than on the day it counts.

Know two numbers

Every rehearsal should leave you with two numbers for each important system:

  1. How long recovery takes. From “it’s gone” to “people are working again”.
  2. How much you would lose. The gap between the last good copy and the moment of loss.

Then ask the business whether those numbers are acceptable. If finance can live without its system for a day but not a week, and the rehearsal took four days, you have found something important — on a quiet Tuesday, with time to fix it.

Put names on it

Finally, make recovery someone’s job. Not a team’s — a person’s, with a deputy. They own the rehearsals, they keep the steps written down, and they are the ones who know where to start when the bad day arrives.

A backup is a copy. A plan is a copy, a set of steps, two numbers the business has agreed to, and a person who has done it before. Only the second one brings you back.