The 3-2-1-1-0 Backup Rule and Immutable Backups Against Ransomware
Nanda Teknoloji Engineering Team · · 8 min read
In brief
The 3-2-1-1-0 rule adds one offline or immutable copy and a goal of zero restore errors to the classic 3-2-1 rule. Because ransomware also targets backups, a copy that cannot be deleted or modified is the last guarantee of recovery.
The 3-2-1-1-0 rule is the classic 3-2-1 backup rule adapted to the ransomware era: keep at least three copies of the data, on two different media, with one copy at another location; in addition, one copy is offline or immutable, and the goal is zero errors in restore verification. The added “1” and “0” matter, because ransomware groups now target not only production data but also backup servers, backup management accounts, and backup repositories. If an attacker can delete the backup, you have no backup. In this article we cover each number in the rule, how an immutable backup works, and common implementation mistakes.
What does the 3-2-1-1-0 rule say, number by number?
| Number | Meaning | What it protects against | Example implementation |
|---|---|---|---|
| 3 | At least three copies of the data (production + two backups) | Corruption of a single copy | Production data, local backup, remote backup |
| 2 | Two different media or storage types | Shared failures in the same technology | Disk-based repository and object storage |
| 1 | One copy at a different location | Site disasters such as fire, flood, or loss of a facility | A copy in another data center or in the cloud |
| 1 | One copy offline or immutable | Ransomware, compromise of a privileged account, malicious deletion | Object storage with an immutability lock, or an air-gapped environment |
| 0 | Zero errors in restore verification | The “the backup was taken but won’t open” surprise | Automatic verification and periodic full restore tests |
The core terms of the rule also appear in the glossary entries for backup and immutable backup.
Why does ransomware attack backups?
Ransomware operators know the victim may recover from backup. That is why, before encryption, these steps are common in the attack chain: obtaining privileged credentials, gaining access to the backup software’s management console, and deleting backup copies and snapshots or shortening retention periods. CISA’s #StopRansomware page also stresses that offline and encrypted backups are among the core measures that must be tested regularly.
Two conclusions follow:
- If the backup sits in the same management and identity plane as production, it goes down together with production.
- Protecting the backup is a job that gets less ownership than taking the backup but is at least as important.
How does an immutable backup work?
An immutable backup is a copy that cannot be modified or deleted for a set retention period after it is written. Common implementation forms are:
- Object storage with object lock. The copy is written with a retention period; until that period ends, no one, including administrators, can delete it.
- Hardened repository. A backup repository protected by an immutability flag in the file system, with management access kept separate.
- Offline media. An environment physically or logically separated from the network and connected only during the backup window.
- Separate identity plane. Backup management is done with accounts that are independent of domain accounts and protected with multi-factor authentication.
Immutability alone is not enough; the retention period must be longer than the time an attacker can remain in the system undetected. Otherwise you only make an immutable copy of data that is already infected.
Replication, snapshot, and backup are not the same thing
These three concepts are often confused and create false confidence.
| Mechanism | Strength | Weakness | Against ransomware |
|---|---|---|---|
| Replication | Low RPO, fast failover | Quickly copies corrupted data too | Insufficient alone |
| Snapshot | Fast and cheap recovery point | Mostly sits on the same storage system | Can be deleted if an administrator is compromised |
| Independent backup | Separate repository, separate retention policy | Restore time can be longer | Strong when combined with immutability |
| Immutable backup | Cannot be deleted or modified | Holds storage for the duration of retention | Last line of defense |
When setting your objectives, the framework in the article on RTO and RPO helps: for a tight RPO, replication and frequent backups; for a safe return to the past, an immutable copy, used together.
How is zero error achieved?
“0” means proving that the backup works when restored. For this, think in three levels:
- Job level: The success status and warnings of every backup job are monitored; failed and skipped jobs are automatically reported to an owner.
- Integrity level: After backup, checksum verification is performed and, where possible, an automatic boot test of the backup in an isolated environment.
- Scenario level: At set intervals, an exercise is run in which a real application, database, and user access are restored end to end. For the method, see the disaster recovery test checklist.
Common mistakes
- Joining the backup server to the domain. If the domain administrator is compromised, the backup goes too.
- Never trying to restore the immutable copy. You will be trying for the first time in a crisis.
- Keeping the retention period too short. If the infection is noticed late, you cannot return to a clean version.
- Backing up only file servers. Identity services, configuration, virtual machine templates, and key management are also part of the recovery chain.
- Keeping encryption keys in the same environment. The keys that open the backup must also be protected and stored separately.
- Leaving the cloud environment on default settings. Object lock and versioning must be explicitly enabled.
Step-by-step path to implementation
- Identify critical systems and their RTO/RPO objectives.
- Lay out in a table which copy, media, and location numbers the existing backups meet.
- Separate backup management accounts from production and require multi-factor authentication.
- Make one copy immutable or offline; set the retention period according to the threat scenario.
- Set up a calendar for automatic verification and periodic restore exercises.
- Write the recovery runbook for the moment of attack, covering who restores what, in which order. We cover the post-attack flow in the article the first 24 hours after ransomware.
Example scenario: When the domain administrator is compromised
Consider the backup infrastructure of a mid-sized organization. The backup software is managed with a domain account, the local disk repository is on the same network, and the remote copy is in a cloud repository managed from the same management console. On paper, 3-2-1 is met: three copies, two media, one remote copy.
The attacker obtains a domain administrator’s credentials through phishing. They log in to the backup console with the same account, delete the local repository, cut the cloud repository’s retention policy to one day and purge the copies, and then encrypt the production systems. The organization had three copies on paper; in practice none remain.
Had 3-2-1-1-0 been applied in the same scenario, the copies in the cloud repository could not have been deleted during the retention period because of object lock, the domain administrator account could not have logged in to the console because backup management was on a separate identity plane, and thanks to regular restore tests the organization would have known in advance which copy it could return from and how long it would take. The difference lies in the last two numbers of the rule.
How are retention period and version count planned?
When setting the immutability period, two questions must be answered together: how long can an attacker stay in the system undetected, and how far back might you need to go to return to a clean point? Short retention periods protect only recently taken backups; if the infection began before that period, all versions may be affected.
A practical approach is a tiered retention policy: frequently taken short-term copies, less frequent medium-term copies, and very infrequent long-term copies are kept together. As tiers increase, storage cost rises too; it therefore makes sense to apply a deeper policy for critical systems and a shallower one for low-criticality systems. Which system falls into which tier comes out of the business impact analysis together with the RTO and RPO objectives.
Checklist: Question your own backup architecture
- Are backup management accounts separate from domain accounts and protected with multi-factor authentication?
- Is at least one of the copies undeletable or offline?
- Is the retention period longer than the time it takes to notice an infection?
- Are the backup software itself and its configuration also backed up?
- Are encryption keys and recovery passwords in a separate, secure place?
- When and on which system was the last restore test done?
- Are identity services, DNS, and configuration data within the backup scope?
Decision summary
- 3-2-1 is the foundation; for ransomware, add an immutable or offline copy for the “1” and verification for the “0”.
- Replication and snapshots do not replace backup; an independent and immutable copy is needed.
- Separate backup management from the production identity plane.
- A backup that has not been restored is an unproven backup; tie exercises to the calendar.
Moving forward with Nanda Teknoloji
On projects like this, Nanda Teknoloji carries out the discovery, design, migration, and operations steps: assessing the existing backup structure against 3-2-1-1-0, designing the immutable copy, and running regular restore tests. For details, see the backup solutions and disaster recovery center pages.