A Successful Backup Job Is Not a Recovery Test
Routine backup software may report success even when copies are incomplete, encryption keys cannot be accessed or a dependent system is missing. A meaningful test restores selected information and verifies that it can support the intended service.
Decide in advance what constitutes success. Recovering a single file does not demonstrate end-to-end restoration for a service dependent on identity, networking and databases.
Select Credible Test Scenarios
Match test cases to critical services, likely failures and approved RTO and RPO targets. Consider accidental deletion, lost infrastructure, ransomware or supplier failure where relevant. Rotate scenarios to cover the important dependencies rather than only convenient samples.
Use isolated recovery environments and appropriate approvals. Tests must not expose sensitive information unnecessarily or overwrite live production data.
Check Integrity and Safe Operation
Verify recovered data completeness, logical consistency and system configuration. In an intrusion scenario, consider whether the restore point itself contains malicious changes. Coordinate with incident responders before reconnecting recovered systems.
Assess the recovered service from a user and security perspective, not just from the backup software console. Record limitations that affect practical service availability.
Measure Results Against Objectives
Record the starting condition, selected recovery point, elapsed time, required staff and manual steps. Compare observed service restoration with approved RTOs and observed data loss with RPOs. Where suppliers participate, document their contributions and evidence.
A failed exercise can be valuable if it leads to tracked corrective action and follow-up testing. Do not replace measured results with a generic statement that the plan was tested.
Maintain Evidence and Review After Changes
Keep test plans, authorisations, results, logs, exceptions and remediation records together. Describe whether the exercise restored one component or an entire service, because those support different assurance claims.
For specified digital and ICT providers, Regulation (EU) 2024/2690 contains further continuity and backup provisions. Other covered entities should assess relevant Member State implementation and competent authority guidance.
Reassess After Material Changes
Cloud migrations, infrastructure redesign, supplier changes and major security incidents may invalidate previous recovery results. Plan additional testing when critical dependencies change.
Document management review of persistent recovery weaknesses and ensure that corrective actions produce demonstrable improvements.
Questions on the text
Does NIS2 impose one annual restoration test for every entity?
The general Directive does not prescribe a single testing frequency for all covered organisations. Check applicable national and provider-specific requirements.
Is recovering files sufficient?
Not when the objective concerns restoring a complete service. Exercise coverage should match the intended recovery outcome.