Guide · AT-R01REV. 2026-A
The 3-2-1 rule: a backup that works when it matters

By Benjamin ReichertNetworks, hardware, support · IT specialist for system integration (Fachinformatiker, IHK)Last updated:
What does the 3-2-1 rule say?
Behind the three digits is a simple idea: any single safeguard can fail, so never rely on one alone.
- Three copies: the original on the server or PC, plus two backups. If one copy fails, two are still left.
- Two media types: Use two different types of storage, such as a backup device on site and encrypted storage in a data centre. A fault that affects one is unlikely to affect the other.
- One copy off-site: against fire, water, burglary and power surges, against anything that would hit the office and the backup at the same time.
Side noteWhy exactly “two media types”?
When the rule was devised, the choice was often between tapes and hard drives. Today the idea is to use two independent systems with separate power, networks and sign-in details. Two USB drives of the same type connected to one server count as a single medium under the rule.
Why is the hard drive next to the server not enough?
The external hard drive is a good start, but it is usually permanently connected to the machine and sits in the same room. Ransomware encrypts everything it can reach, and that includes the plugged-in backup. A fire or a burglary hits the original and the copy at the same moment. The hard drive check shows the state of the drive itself.
A power cut can also cause problems. “Power cuts at work” explains what happens when a NAS or server suddenly loses power and which uninterruptible power supply can keep it running.
A synchronised cloud folder is not a backup either. The big services now offer versions and a recycle bin, but the scope of the backup is not defined anywhere: databases, open files and the system state are left out. There is also no proof that anything can be restored from it. And the whole thing hangs on one account and licence. Whoever takes over that account takes over the supposed backup along with it.
The part that really counts: restoring
In practice, backups rarely fail at backing up. They fail at restoring. The backup reported “successful” for months but was incomplete. Nobody remembers the password for the encryption any more. Or restoring does work, but takes days while the business stands still.
That is why every backup needs a restore test: regularly bring back a single file, and at longer intervals a complete system, measuring how long it takes. Only this record turns hope into proof. A backup that has never been restored is not a safeguard but a promise.
Side noteTwo questions your emergency plan must answer
First: how much work can you afford to lose at most, a day or an hour? That decides how often you back up. Second: how quickly must you be able to work again? That decides the technology. Specialists call these RPO and RTO; more important than the abbreviations is that the answers are written down.
How to put the rule into practice in your business
Four things have proved their worth. First: back up automatically, because anything that has to be started by hand will be skipped sooner or later. Second: the backup reports by itself when something goes wrong, instead of failing silently. Third: at least one copy is disconnected from the network or stored in a form that cannot be changed, so that an attacker cannot hit it as well. Fourth: an emergency plan is ready in plain language, covering who does what, in which order, with which login details.
This is exactly how we set up data backup for our customers: following the 3-2-1 rule, with a verified restore and an emergency plan that non-technical people can understand. More under Backup & disaster recovery. So that a server failure stays a bad morning rather than ruin.
The IT emergency check shows where your business stands in a few minutes. It asks about the four steps above as well as power, access and documentation, then estimates how long it would take to get back up and running.