Skip to main content
Legia

IT Infrastructure & Cloud

The 3-2-1 Backup Strategy for Company Servers

The 3-2-1 rule is the best-known backup pattern: three copies of data, on two types of media, with one copy offsite. This guide applies it to company servers, including lessons from ransomware attacks.

SEO and Content Specialist

4 min read

A patch panel with blue and gray network cables

Short answer: the 3-2-1 rule means keeping three copies of important data (one primary and two backups), on two different types of media, with one copy offsite. For company servers today, add two more things: one copy that cannot be changed or deleted from the main network, and regular recovery tests. A backup that has never been tested may not be recoverable.

The value of a backup is only felt when it is needed: a server fails, data is deleted by accident, or ransomware encrypts everything. At that moment the question is no longer “do we have a backup”, but “is the backup complete, intact, and recoverable within a reasonable time”. The 3-2-1 rule helps answer that question from the start.

The 3-2-1 rule in one table

NumberMeaningProtects against
3 copiesOne primary data set and two backupsOne backup turning out to be damaged or incomplete
2 mediaBackups on two different types of storage, for example a network storage device and a cloud serviceFailure of one type of media
1 offsiteOne copy in a place other than the main office or data centerFire, flood, theft, or disruption at one location

This rule is described, among other places, in the “Data Backup Options” document published by US-CERT (now CISA).

Illustration: server

Adjustments for the ransomware threat

Modern ransomware does not only encrypt the primary data, but also looks for and damages backups reachable from the same network. For that reason, the 3-2-1 pattern for company servers should be extended with:

  • One immutable copy, or one disconnected from the network (offline) for a set period, so it cannot be deleted by an account taken over by an attacker.
  • Separate credentials for the backup system. The everyday server admin account should not be able to delete backups.
  • History across several points in time, not only the latest copy. If an infection is not detected right away, the latest copy may already be infected.

The importance of separate backups was clear in the 2024 Temporary National Data Center incident, discussed in the article Ransomware in the Office: Lessons from the 2024 National Data Center Incident.

What needs to be backed up

A server backup is not only files. Make sure the list covers:

  1. Data: application databases, shared files, and attachments.
  2. Configuration: settings for servers, applications, and network devices.
  3. Systems: a server image if fast recovery is needed, or at least notes on how to rebuild it.
  4. Credentials and keys needed for recovery, stored securely and separately.

Deciding how often and how long

Two questions determine the backup schedule:

  • How much data can be lost? If losing a day of data is acceptable, a daily backup is enough. For transaction systems, the interval between backups needs to be much shorter.
  • How long can the system be down? This determines the recovery method: restoring files one by one, or restoring the whole server from an image.

Also decide how long each backup is kept, taking into account operational needs and the company’s data retention policy.

Recovery testing: the part most often skipped

A simple recovery test can be run regularly: pick one system, restore it from backup into a separate environment, check whether the application runs and the data is intact, and record how long it took. That record becomes a realistic basis for the recovery plan. Many problems are only discovered through a test like this: backups that turn out not to contain the database, encryption keys that were not saved, or a recovery that takes days.

Example setup for a mid-sized office

As an illustration, an office with one file server and one application server could apply the following pattern:

CopyLocationNotes
Primary dataServer in the officeUsed day to day
First backupNetwork storage device in another roomFast recovery for deleted files
Second backupCloud storage with deletion protectionOffsite, safe from fire and ransomware

This example is an illustration, not a design for any particular office. Schedules, retention periods, and media types are adjusted to the needs of each system.

Common mistakes

  • Backups on the same disk as the primary data, or on a device that is always connected to the network.
  • Only the latest copy, with no history across points in time.
  • No one monitors whether the backup process succeeds every time it runs.
  • Treating synchronization as a backup. Files deleted or encrypted in one place are synchronized to the other place too.

How Legia does it

In the Server Installation & Management service, backups are part of the installation and are tested at installation. Regular recovery tests can be run if agreed in the management contract. The installation also covers a firewall, SSH keys, two-step authentication, and access logs, for office servers, virtual servers, and cloud alike. Considerations for choosing where to host a server are in the article Office Server, Cloud, or Shared Hosting.

Frequently asked questions

Does a cloud backup satisfy “one offsite”?

It can, as long as that copy is not deleted along with it if the main system or the main admin account is compromised. Access permissions and deletion protection still need to be set up.

Does syncing files to the cloud count as a backup?

No. Synchronization copies changes, including deletions and files encrypted by ransomware. A backup keeps points in time that can be restored.

How often should recovery tests be done?

According to how important the system is and how often it changes. Most importantly, tests are done regularly and the results recorded, not just once at installation.

Sources

Category IT Infrastructure & Cloud

Keep reading

Related articles

All Articles