A backup that has never been restored is an assumption, not a control. EU GMP Annex 11 asks for more than a nightly job: data must be protected from loss and remain accessible, readable and intact for the whole retention period.
What Annex 11 actually says
Annex 11 expects regular backups of all relevant data, with the integrity and accuracy of backup data and the ability to restore them checked during validation and monitored periodically. For archived data, it expects checks of accessibility, readability and integrity, and, if relevant changes are made to the system, assurance that data can still be retrieved and that this has been tested.
Backup and archive are different things
A backup exists to recover a system after failure. An archive exists to preserve records for the retention period, often after the system has been retired. They have different owners, media and tests, and mixing them up is a common finding. Define for each system which data is covered, how long it is kept, and who is responsible.
Design a restore test that means something
A good restore test does more than show that files reappear. Consider:
- Restoring to a separate environment and confirming that records open, calculations run and audit trails are present
- Including metadata and audit trails, not only the final results
- Testing the restore time against the recovery objective in your business continuity plan
- Repeating the test after major upgrades, storage changes or migration to a new platform
- Documenting the result and any deviation, with the frequency set by risk
Protect the archive from obsolescence
Readable today does not mean readable in ten years. Proprietary formats, retired applications and ageing media all threaten access. Options include maintaining a read-only instance, converting to a stable format while preserving context and audit trails, or retaining the application in a controlled way. Whichever you choose, verify periodically that records can be found and read, and record the outcome.
Know where the data lives
Cloud services, local instrument PCs and shared drives are frequent blind spots. Include them in the system inventory, and check that quality agreements with providers clarify who performs and tests backup and how you can obtain your data on exit.
A pragmatic way to begin
If the topic feels large, begin with the systems that influence release decisions, patient safety or regulatory submissions. Confirm they are on the inventory, check who has access, check that a restore or recovery has been tested, and check that the audit trail is reviewed in practice. These four questions take little time and usually show where effort should go first.
How PQRA helps
PQRA can review your backup, restore and archiving arrangements against Annex 11 and data integrity expectations, help design risk-based restore tests, and support archiving and decommissioning plans for legacy and retired systems.
Contact PQRA to discuss your computerised systems and data governance programme.


No responses yet