Backup strategy: the 3-2-1 rule, ransomware protection, restore tests and a template
A backup strategy decides what data is copied, how often, where the copies are kept, how long they are kept, and how quickly a system can be restored from them. Backups are the last line of defence against ransomware, deleted data, failed updates and outages, and they only count if a restore has actually been tested. This guide covers what to back up, backup types, the 3-2-1 rule and its modern additions, recovery objectives, restore testing, the legal requirements, and a template outline for a written backup strategy.
Sources were read on 12 November 2026: Directive (EU) 2022/2555 (NIS2), Regulation (EU) 2016/679 (GDPR), the CISA and MS-ISAC #StopRansomware Guide, and the 2012 paper "Data Backup Options" produced for US-CERT.
What a backup strategy must answer
- What: which systems and data, including the ones people forget.
- How often: the frequency per system, derived from how much data loss is acceptable.
- Where: storage locations, media and regions, and how they are separated from production.
- How long: the retention of daily, weekly, monthly and yearly copies.
- How fast back: the time to restore each system, and in which order.
- Who: who runs, monitors and tests backups, and who can delete them.
- Proof: how restore tests are run and recorded.
What to back up
Databases and file servers are obvious. The gaps are usually elsewhere: SaaS data such as email, documents, CRM and accounting in the cloud; configuration of firewalls, network devices and cloud accounts; code repositories and build pipelines; secrets and encryption keys, kept safely and separately; virtual machine images or, better, the infrastructure-as-code definitions that rebuild them; and laptops where people keep work outside shared storage. Make an inventory of systems first, then decide per system. If a system can be rebuilt from code and configuration, back up those and the data, not the whole machine.
Backup types: full, incremental, differential
| Type | What is copied | Restore needs | Trade-off |
|---|---|---|---|
| Full | Everything, every time | The one full backup | Simplest restore, most storage and time |
| Incremental | Changes since the last backup of any type | Last full plus every incremental since | Fast and small, longer restore chain |
| Differential | Changes since the last full backup | Last full plus the latest differential | Middle ground between the two |
| Snapshot | A point-in-time image of a disk or database | The snapshot | Fast, but often stored with production, so not a backup on its own |
| Continuous / point-in-time | Every change, e.g. database transaction logs | Base backup plus logs to a chosen moment | Minimal data loss, needs managed tooling |
Retention is often organised in generations: daily copies kept for a few weeks, weekly copies for a few months, monthly copies for a year or more. The schedule must also respect your data retention policy: data deleted under the policy should disappear from backups within a known time.
The 3-2-1 rule and protection against ransomware
The 2012 paper "Data Backup Options", produced for US-CERT, set out the 3-2-1 rule: keep 3 copies of any important file (1 primary and 2 backups), keep the files on 2 different media types to protect against different types of hazards, and store 1 copy offsite. The rule is still the foundation, but ransomware changed what "offsite" must mean.
The CISA and MS-ISAC #StopRansomware Guide advises maintaining offline, encrypted backups of critical data and regularly testing their availability and integrity in a disaster recovery scenario. It explains why offline matters: many ransomware variants attempt to find and subsequently delete or encrypt accessible backups to make restoration impossible unless the ransom is paid. It also recommends using infrastructure as code to deploy cloud resources and keeping backups of template files offline to redeploy quickly. In practice, a modern strategy adds to 3-2-1:
- One copy offline or immutable, using object lock, write-once storage or a backup service with locked retention.
- Separate credentials: the backup system is not administered with the same accounts as production, and deletion needs extra approval.
- Encryption of backups at rest and in transit, with keys stored separately from the backups.
- Monitoring of failed jobs and of unusual deletion or change rates.
Recovery objectives: RPO and RTO
Two numbers turn a backup strategy from guesswork into a plan. The recovery point objective (RPO) is the maximum acceptable data loss, measured in time: an RPO of one hour means backups must be at most one hour apart. The recovery time objective (RTO) is the maximum acceptable time until the system works again. Both are business decisions, made per system with the people who depend on it, and they usually come from the business impact analysis in the business continuity plan. A webshop's order database may need an RPO of minutes and an RTO of hours; an archive of old projects may accept a week for both.
Restore tests: the only proof
A backup that has never been restored is a hope. Restore tests should be planned, varied and recorded:
- Single file or record restores, monthly, the most common real need.
- Full system restore to a separate environment, at least every quarter for critical systems.
- Disaster scenario, yearly: assume production and its accounts are gone, and rebuild from the offline or immutable copy alone.
- Measure and record the time taken, problems found and whether the RTO and RPO were met.
Legal requirements
Under the GDPR, Article 32(1) requires security appropriate to the risk, including, as appropriate, "the ability to restore the availability and access to personal data in a timely manner in the event of a physical or technical incident" and a process for regularly testing, assessing and evaluating the effectiveness of the measures. For essential and important entities, NIS2 Article 21(2)(c) lists "business continuity, such as backup management and disaster recovery, and crisis management" among the minimum risk-management measures. Backups are also part of the information security policy and are central to an incident response plan for ransomware.
Backups for your own software and apps
For custom software, the backup strategy is part of the architecture. Managed databases with point-in-time recovery, object storage with versioning and object lock, a second region or provider for the immutable copy, and infrastructure as code so the whole environment can be recreated from the repository. User-uploaded files need the same protection as the database, and a restore must bring both back to the same point in time, or records will point to files that do not exist.
A backup strategy template outline
- Purpose, scope, owner, version, review date.
- System inventory with data owner, RPO and RTO per system.
- Backup method per system: type, frequency, tool, location.
- Retention schedule: daily, weekly, monthly, yearly copies, aligned with the retention policy.
- Protection: offline or immutable copy, encryption, key management, access rights.
- Monitoring and alerting: who is told when a job fails.
- Restore procedures per system, step by step, stored where they can be reached when systems are down.
- Restore test plan and test log.
- Roles and contacts, including suppliers.
How we help
In the systems we build, backups with an immutable copy, point-in-time database recovery and infrastructure as code are part of the delivery, documented with restore procedures and tested before go-live. Our managed IT services include backup monitoring and scheduled restore tests with recorded results, and our software maintenance keeps backup jobs and restore procedures current as systems change. Backups stay in EU regions, in accounts in your company's name.
Questions people ask
What is the 3-2-1 backup rule?
Keep three copies of important data (the original and two backups), on two different types of storage, with one copy off site. A guide produced for US-CERT in 2012 describes it in exactly those terms. Today most companies add that one copy should be offline or immutable, so ransomware cannot reach it.
How often should we back up?
As often as the business can afford to lose data. That is the recovery point objective: if losing one day of orders is acceptable, a nightly backup may do; if losing one hour is not, you need continuous or hourly backups of that system. Different systems usually need different frequencies.
Are cloud services like Microsoft 365 or Google Workspace backed up automatically?
The provider protects its infrastructure against its own failures, but deleted or encrypted data, a compromised account or a sync that overwrites files may still be your problem to recover. Check the provider's retention and recovery features and, for important data, add a separate backup to another service.
What is an immutable backup?
A backup that cannot be changed or deleted for a set period, even by an administrator account. Object storage with object lock, write-once storage and backup services with locked retention provide it. It protects against ransomware and against an attacker who has gained admin access.
How do we know our backups work?
Only by restoring them. Schedule restore tests, at least every quarter for critical systems and after major changes, restore to a separate environment, check that the data and application work, record how long it took, and compare with the recovery time objective.
Sources
- CISA and MS-ISAC, #StopRansomware Guide (read 12 November 2026)
- Paul Ruggiero and Matthew A. Heckathorn, Data Backup Options, produced for US-CERT, 2012 (PDF) (read 12 November 2026)
- Regulation (EU) 2016/679 (GDPR), Article 32(1) (read 12 November 2026)
- Directive (EU) 2022/2555 (NIS2), Article 21(2)(c) (read 12 November 2026)