An example is a backup copy stored on media or in a location that is not continuously connected to production systems. Practitioners use it to preserve a recovery point that malware cannot easily encrypt or delete. Test restore access, protect backup credentials, and ensure the gap is real, not just logical segmentation.
What an air-gapped backup looks like in practice
An air-gapped backup is usually a copy of data that cannot be reached from the production network during normal operations. In practice, that can mean removable media that is taken offline after the backup job, or a backup repository separated by strict access and connectivity controls so ransomware cannot touch it from an infected system.
The key operational idea is not just “backup exists”, but “backup is unreachable when production is compromised”. That is why a genuine air gap is often paired with immutable storage and separate administrative access, as explained in the Backup Security and Immutable Backup Guide.
A practical example might be a weekly full backup written to encrypted tape, then removed from the library and stored offsite, with the backup catalog retained separately so the tapes cannot be browsed or deleted from the primary environment. Another common pattern is a dedicated recovery vault that is only mounted during backup windows and otherwise remains isolated from production credentials and traffic.
How practitioners preserve the gap without losing recoverability
Air-gapped backup only works if the separation is operationally real. Teams need to verify that production admin accounts, backup operators, and service accounts cannot directly modify the offline copy, and that any management plane used to bring backups online is itself tightly controlled. Restore testing matters because an isolated backup that cannot be restored quickly is only theoretical resilience.
That is why restore drills should confirm both data integrity and access path integrity: who can mount the backup, how the data is decrypted, and whether the recovery environment is clean enough to trust. The recovery workflow in the Ransomware Response and Recovery Guide is relevant here because backup recovery is part of the broader containment and restoration sequence.
Good practice is to treat the backup system as a separate trust domain. If an attacker can reach the backup console, reuse the same credentials, or delete snapshots through the same management path, then the “air gap” has collapsed into a normal online backup with different branding.
Where air-gapped backups fail in real environments
The most common failure is a logical gap that is mistaken for a physical one. A backup may be in a separate subnet, account, or storage bucket, yet still be continuously reachable by compromised automation, shared credentials, or privileged APIs. That creates enough access for ransomware to encrypt, erase, or corrupt recovery points before anyone notices.
Failure mechanism: The backup copy is still addressable through live credentials or a mounted control plane, so malware or an intruder can reach it even though the design appears segmented.
Impact: Recovery becomes dependent on the same compromised environment that caused the incident, which can leave the organisation with no trustworthy restore point when it needs one most.
Another recurring issue is overconfidence in retention settings. Long retention alone does not protect a backup that can be deleted, and immutability alone does not help if the backup was already poisoned before the immutable window began. Air-gapped storage works best when it is combined with separation of duties, periodic validation, and a clean recovery process.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
CIS Controls v8 and NIST SP 800-53 Rev 5 set the technical controls, while ISO/IEC 27001:2022 defines the regulatory obligations.
| Framework | Control / Reference | Relevance |
|---|---|---|
| CIS Controls v8 | CIS-10 — Data Recovery | Air-gapped backups are a recovery safeguard that limits ransomware impact. |
| Recommendation — Protect offline recovery points and test restore capability on a regular cadence. | ||
| NIST SP 800-53 Rev 5 | CP-9 — System Backup | The topic is directly about backup copies and recoverability under compromise. |
| CP-10 — System Recovery and Reconstitution | Air-gapped backups support recovery after destructive malware or outage events. | |
| IA-5 — Authenticator Management | Backup systems depend on protected credentials that must not expose recovery copies. | |
| Recommendation — Maintain protected backups and verify they can be restored when needed. Define and test recovery procedures that use isolated backup copies. Rotate and protect backup credentials so they cannot be reused to reach recovery media. | ||
| ISO/IEC 27001:2022 | A.8.13 — Information backup | Air-gapped backups are a direct implementation of protected backup storage and recovery. |
| Recommendation — Store backup copies so they remain available when primary systems are compromised. | ||
Practitioner Guidance
What to verify: Confirm that the backup copy is truly offline or otherwise unreachable from production except during a controlled recovery or backup window. Test whether production credentials, backup admin accounts, and automation tokens can alter or delete the copy; if they can, the gap is not strong enough.
What good looks like: A valid air-gapped design has a documented way to create backups, a separate way to access them, and a restore process that has been exercised on clean systems. The backup should survive a full production compromise without relying on the same identity stack or management plane.
Common mistake: Treating network segmentation as equivalent to an air gap. If a backup remains continuously mounted, API-reachable, or editable by the same operational credentials, it is still exposed to the same blast radius as the primary environment.
Practitioner takeaway: The point of an air-gapped backup is not just preservation, it is survivable recovery after the production environment is no longer trustworthy.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Free weekly newsletter
Subscribe to the NHI & AI Identity Journal
The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.
Bonus 33% off our NHI Course when you subscribe.
Reviewed and updated by the NHIMG editorial team on October 19, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org