The 3‑2‑1 backup rule, explained
Three copies, two types of storage, one copy off-site. Why this simple rule still works so well — and how to apply it in practice.
Nebora Cloud engineering team
Most data loss does not start with a disaster. It starts with a deleted folder, a failed disk, a bad update or ransomware that quietly encrypts a file share over the weekend. The 3-2-1 rule is the simplest way to make sure none of these turns into a business problem.
What the rule says
- 3 copies of your data: the production data plus two backups.
- 2 different types of storage: for example, a backup appliance and object storage, so one fault cannot destroy both.
- 1 copy off-site: physically separate from the systems it protects.
The idea is not new, but it has aged well. Each part removes a different single point of failure.
Why three copies
A single backup is a single point of failure. Backups fail more often than people expect: a job silently stops after a password change, a disk fills up, or the only copy is corrupted at the moment you need it. A second backup makes the chance of losing everything at once very small.
Why two types of storage
If both backups sit on the same storage system, a firmware bug, a controller failure or one compromised admin account can affect both. Using two technologies — or at least two independent systems — means a problem in one does not spread to the other.
Why one copy off-site
Fire, flooding and power events affect a whole building. So does ransomware that reaches every server on the same network. An off-site copy, ideally immutable (it cannot be changed or deleted for a set period), is the copy that saves you when everything else is affected.
Going one step further: 3-2-1-1-0
Many teams now extend the rule:
- 1 copy offline or immutable, so it cannot be encrypted by an attacker who gains admin access.
- 0 errors after restore testing — a backup only counts if you have proven you can restore it.
How to apply it in practice
- List what matters. Databases, file shares, mailboxes, virtual machines and SaaS data such as Microsoft 365.
- Set RPO and RTO for each. How much data can you afford to lose (RPO), and how quickly must it be back (RTO)?
- Automate the schedule. Daily backups at minimum; more often for critical databases.
- Encrypt everything. In transit and at rest, with keys you control.
- Test restores every month. Restore a file, a database and a full server, and write down how long it took.
How we handle it at Nebora
Our Backup service keeps encrypted copies on separate storage in our Sofia data center, with an optional immutable copy and monthly restore tests reported in the Client Portal. Your data stays in the EU, and our engineers check failed jobs before you notice them.
If you are not sure your current setup meets the 3-2-1 rule, we are happy to review it with you.