Backups and Disaster Recovery: The 3-2-1 Rule Explained
Most people who lose important data did technically have a backup — it just wasn't a good enough one. The 3-2-1 rule is the industry's answer to "how many backups is actually enough," and it is simple enough to remember under pressure.
Updated 2026-08-06
The rule in one sentence
Keep three total copies of your data, on two different types of storage media, with one of those copies stored somewhere physically separate. Each number in that sentence is defeating a different way backups commonly fail.
Why three copies, not one
A single backup is a single point of failure — if that one backup drive fails at the same moment as your original, you have nothing. Three total copies (the original plus two backups) means two independent failures would need to happen at once before you actually lose the data.
Why two different media types
If both your original and your backup live on the same type of storage — two hard drives from the same batch, for instance — they can share the same weaknesses and fail for the same reason at a similar time. Spreading copies across genuinely different media (a local drive and a cloud service, for example) means a flaw or failure specific to one technology doesn't take out both copies at once.
Why one copy has to leave the building
This is the part people skip, and it's the part that matters most for the worst-case scenarios: fire, flood, theft, or a natural disaster that destroys everything in one physical location at once. An offsite copy — a cloud backup, or a drive kept somewhere else entirely — survives events that would wipe out every local copy simultaneously.
Ransomware changed the calculation
Modern ransomware actively searches for and encrypts connected backup drives too. An offsite or disconnected backup isn't just protection against fire and theft anymore — it may be the only copy ransomware cannot reach at all.
A backup you have never tested is a guess, not a backup
The final, often-skipped step is restoring from a backup occasionally just to confirm it actually works. A backup process that silently stopped running months ago, or that produces files nobody can actually restore from, provides false confidence right up until the moment it matters — which is exactly the wrong moment to discover the problem.