What Are Air Gapped Backups? Examples & Best Practices

Air Gapped Backups keep at least one recovery copy beyond attackers' reach. Learn how they work, types, examples, and best practices for ransomware recovery.

What Are Air Gapped Backups?

Ransomware operators now target backup infrastructure before encrypting production data. According to CISA's cyber incident cost study, the mean cost per breach incident reaches $5.9 million, which is why your last line of defense is only useful if attackers cannot reach it.

Air gapped backups are isolated copies of critical data that are physically or logically separated from production networks. They create a protective barrier by establishing environments with controlled access that network-based attacks cannot reach. Whether through physically disconnected tape media, network-segmented storage with strict access controls, or logically isolated cloud vaults, air gapped backups help ensure at least one recovery copy remains harder for attackers to reach when they move laterally through your environment.

The NIST SP 800-209 establishes that backup operations may include storage devices that are offline, while the CISA ransomware guide directs organizations to maintain offline backups because ransomware often attempts to delete or encrypt accessible backup data.

Air Gapped Backups as a Cybersecurity Control

Air gapped backups sit at the intersection of data protection and ransomware resilience. Modern ransomware families conduct reconnaissance specifically to locate backup servers, delete shadow copies, and encrypt backup repositories before triggering production encryption. The CISA ransomware response guide directs organizations to "maintain offline, encrypted backups of critical data" because modern ransomware variants "attempt to find and subsequently delete or encrypt accessible backups."

When your endpoint protection finds and stops a ransomware attempt through Behavioral AI, the threat ends there. But when defenses are defeated or bypassed, air gapped backups can provide a recovery path outside the immediate blast radius of a network-based attack. To understand how that protection works in practice, you need to break the architecture into its core components.

Core Components of Air Gapped Backups

Every air gapped backup architecture, regardless of type, is built from the same six layers. Each one must hold for the isolation to be meaningful.

• Isolated Storage Layer. The physical or logical environment where backup copies reside, including tape libraries, removable disk arrays, network-segmented systems, or isolated cloud security domains with separate authentication.
• Transfer Mechanisms. Data moves from production to the isolated storage layer through controlled pathways, whether manual or robotic media transport, or scheduled replication with one-way data flows that prevent reverse contamination.
• Immutability Layer. Once data reaches the air gapped environment, it should resist modification or deletion through hardware write-once tape media, software retention locks, or object storage compliance modes.
• Access Controls. Physical implementations depend on key or combination access with logged entry. Logical implementations require role-based access control, multi-factor authentication, least-privilege enforcement, and audit logging. The CISA data protection guide requires MFA for administrative access and role-based access with least-privilege principles across protected systems.
• Encryption. Backup data should be encrypted at rest and in transit. The CISA encryption standards guide specifies FIPS 140-2 for data in transit and at rest, or enhanced encryption mechanisms. Encryption key management systems should remain isolated from backup data to prevent a single compromise from exposing both.
• Verification Infrastructure. You need the ability to confirm backup integrity through regular integrity checks, checksums, backup verification routines, and recovery testing.

Weakness in any single layer can undermine the entire architecture. The next step is understanding how these layers operate across backup, isolation, storage, and recovery.

How Air Gapped Backups Work

Air gapped backups follow a four-phase cycle: capture, isolation, secure storage, and controlled recovery. Each phase has specific requirements that determine how well the architecture holds when an attacker reaches your environment.

• Phase 1: Backup and Capture. Production data is copied to the backup target through your standard backup processes. Before data enters the air gapped environment, anti-malware scanning validates integrity. Pre-gap scanning is essential because backing up compromised data into an immutable, air gapped environment creates a poisoned copy that could trigger reinfection during recovery.
• Phase 2: Isolation. For physical air gaps, this means removing tape cartridges or disconnecting removable disk arrays from all systems and networks. For logical air gaps, network connectivity is disabled between backup cycles, with scheduled replication windows requiring explicit authentication per transfer. Cloud-based vaults enforce isolation through separate security domains with API-only access and independent authentication.
• Phase 3: Secure Storage. Isolated copies remain in their protected state, with physical media stored in secure off-site vaults and logical copies maintained behind network segmentation, access controls, and immutability locks. This backup isolation is what keeps at least one recovery path less exposed when attackers move through connected systems.
• Phase 4: Recovery. When you need the data, the process reverses. Physical media is retrieved from secure storage and manually connected. Logical copies are accessed through authenticated, controlled pathways. Restore to an isolated staging environment for verification before reconnecting to production.

Once you understand the workflow, you can evaluate which implementation model fits your environment best.

Types of Air Gapped Backup Architectures

You have three primary architecture choices, each with distinct trade-offs.

1. Physical Air Gaps use removable storage media with complete network disconnection. Tape cartridges are copied, physically removed, and stored off-site. This approach offers low storage cost per gigabyte and the strongest ransomware isolation. The trade-off is slow data access and unsuitability for organizations requiring rapid, frequent restores.
2. Logical Air Gaps use network segmentation, protocol restrictions, and access controls to create isolation without physically removing media. Components include separate VLANs with firewall controls, elimination of commonly targeted protocols such as CIFS, NFS, and SMB, and one-way data flows. Logical air gaps suit enterprise-scale environments where physical disconnection is impractical. This model depends on strong identity, segmentation, and policy enforcement.
3. Cloud-Based Air Gapped Vaults create logically isolated security domains within cloud infrastructure. These use separate authentication, object-level immutability such as S3 Object Lock, API-only access, and MFA. Cloud storage alone does not constitute an air gap; additional isolation controls are required. If your environment spans cloud infrastructure and cloud application backups, your cloud security controls shape how strong that logical separation really is.

The architecture choice sets your recovery constraints. The examples below show what each looks like in a real environment.

Air Gapped Backup Examples in Practice

Understanding the architecture in the abstract is one thing. Seeing it in practice makes implementation decisions clearer. Here are three scenarios showing what air gapped backups look like across different environments.

Tape-Based Physical Air Gap

A manufacturer running industrial control systems backs up critical OT configurations and production historian data nightly to LTO tape. After each job completes, a technician removes the cartridge, logs it in a chain-of-custody register, and stores it in a locked fireproof safe off the production floor. The tape maintains no network connection at any point. When recovery is needed, the tape is retrieved, connected to an isolated workstation, scanned for integrity, then used to restore. The CISA ICS security guidance identifies offline media storage as a baseline control for operational technology environments.

Network-Segmented Logical Air Gap

An enterprise runs backup software on a hardened server sitting on a dedicated backup VLAN, separated from all production segments by firewall policy. SMB, NFS, and CIFS are disabled on that segment. During scheduled replication windows, a one-way data flow copies backup data: each job requires MFA-protected authentication with a service account holding no other network privileges. No domain-joined production endpoint can reach the backup server directly, cutting off the lateral movement paths ransomware uses to locate and destroy backup infrastructure.

Cloud Vault Logical Air Gap

A cloud-native company stores backups in an AWS S3 bucket with Object Lock enabled in compliance mode, inside a separate AWS account isolated from the production account. No IAM role in the production environment holds write or delete permissions on the backup bucket. Backup jobs run via a one-way API call authenticated with a dedicated credential set that exists only in the backup account. Even with full production account compromise, an attacker cannot delete or overwrite vault contents during the retention period. This model aligns with CISA ransomware guidance on maintaining backups behind separate credentials and access controls.

Each of these implementations fulfills the offline copy requirement in the modern 3-2-1-1-0 backup rule: one copy that is offline, immutable, or air gapped, with zero errors confirmed through restore testing rather than checksums alone. That framework points directly to the core benefits air gapped backups provide.

Key Benefits of Air Gapped Backups

When implemented correctly, air gapped backups deliver four security and operational advantages that no always-connected backup architecture can match.

• Ransomware Isolation. This is the primary value proposition. Air gapped backups eliminate the network-accessible pathways that enable backup destruction. Physical or logical isolation means ransomware executing on production systems cannot easily reach the backup copy.
• Recovery Confidence. Air gapped backups can preserve recovery points that are less exposed to active attacks. Organizations with compromised backups must conduct forensic analysis to determine which backup generations are trustworthy, a process that can significantly extend recovery. Air gapped architectures help reduce that pressure by limiting direct attacker access to at least one recovery path.
• Regulatory Alignment. Air gapped backup strategies align with the NIST SP 800-209, support contingency planning expectations in federal security and healthcare environments, and generally support backup, availability, and access-control objectives found across major control frameworks. For regulated industries, air gapped architectures can provide defensible evidence of reasonable cybersecurity controls.
• Insider Threat Reduction. Air gapped architectures shrink the access surface by design. Even privileged users cannot access backup repositories through standard network pathways, requiring physical access or dual-approval workflows for any interaction with isolated copies. This is the recovery-side control that keeps a single compromise from becoming a business-ending event.

These benefits are real, but they raise an equally important question: which organizations actually need air gapped backups, and at what level?

Who Needs Air Gapped Backups

The question is not whether your organization is large enough; it is whether you can afford to lose production data access with no verified, attacker-resistant recovery copy. The table below maps organization type to the most practical air gap approach and its primary driver.

Organization type	Recommended approach	Primary driver
Critical infrastructure (energy, utilities, OT)	Physical tape air gap	Nation-state threat exposure, regulatory mandate
Healthcare networks	Logical air gap or cloud vault	HIPAA contingency planning, patient data recovery
Financial services	Logical air gap with immutability	FFIEC, PCI DSS compliance, tight RTO requirements
Mid-market enterprises	Cloud vault with Object Lock	Cost efficiency, limited on-site storage capacity
Cloud-native and SaaS companies	Cloud vault in a separate account	No on-premise infrastructure, production reachability risk
Government agencies	Physical tape (FIPS-compliant encryption)	FISMA, NIST SP 800-53 contingency planning

Smaller organizations often assume air gapped backups are operationally out of reach. In practice, a cloud vault with Object Lock in a separate account requires no on-premise hardware and can be configured in hours. A ten-person medical practice has just as much recovery leverage to protect as a large enterprise; the ransomware arithmetic is the same regardless of headcount.

Knowing which approach fits your environment is essential, but air gapped backups also come with real trade-offs. Understanding them before deployment prevents architectural decisions you'll need to reverse under pressure.

Challenges and Limitations of Air Gapped Backups

Air gapped backups are not a turnkey solution. Four challenges consistently cause teams to either misconfigure their architecture or overestimate the protection it actually provides.

• Definitional Ambiguity Creates False Security. Most "air gapped" backup deployments are not truly air gapped. An authentic air gap requires systems that are not connected physically and where any logical connection is not autonomous but controlled manually. You need to know exactly which type you have deployed.
• The Automation vs. True Isolation Paradox. Enterprise environments face a fundamental tension: true air gapping requires manual intervention, but manual processes at scale create prohibitive costs. Any autonomous pathway, whether scheduled rsync jobs, API calls, or backup agents, creates exploitation opportunities. This tension cannot be fully resolved, only managed through deliberate architectural decisions.
• Operational Complexity and Extended Recovery Times. Air gapped implementations create operational friction that directly impacts recovery objectives. The increased complexity of backup and recovery processes can lead to longer recovery times compared to always-connected solutions, a paradox where enhanced security slows recovery during the exact incidents the architecture was designed for.
• Immutability Is Not Air Gapping. Organizations frequently conflate these two distinct controls. An immutable backup that contains ransomware or malware is useless during recovery and could trigger reinfection. An air gapped backup that was never scanned may contain corrupted data. Both controls are needed, with a clear understanding of what each addresses. The most common implementation errors appear when teams blur those distinctions in day-to-day operations.

That is why it helps to review the mistakes that repeatedly weaken otherwise sound designs.

Common Air Gapped Backup Mistakes

Even well-designed air gap architectures fail in practice when the same operational errors go uncorrected. These six mistakes account for the majority of backup environments that appear isolated but remain reachable.

• Calling logical isolation "air gapped" without documenting the accepted risk. When you implement network-segmented backup systems with scheduled transfers, you have logical isolation, not a true air gap. Document the accepted risk and compensate with additional layers: MFA, RBAC, immutable storage with retention exceeding typical attacker dwell times, and anomaly monitoring.
• Leaving backup infrastructure on the production network. Allowing backup systems to remain accessible from the same network segments as production systems is a significant architectural error. Segregate backup infrastructure onto dedicated segments. Management interfaces should never be accessible from general corporate networks.
• Ignoring default credentials on backup software. Some backup software still ships with default user logins and passwords. Combined with a lack of MFA, this gives attackers a direct entry point. Eliminate defaults, enforce MFA for all backup access, and implement dual-approval workflows for destructive operations like deleting backup data before its scheduled expiration.
• Never testing restores. This is the single most common gap. The CISA ransomware response guide emphasizes testing backup procedures regularly. You need to test full system restoration from air gapped media, not just verify file integrity. Schedule regular full restore tests to isolated environments and measure actual recovery times against your Recovery Time Objective (RTO) requirements.
• Skipping pre-gap malware scanning. Backing up compromised data into an immutable, air gapped environment creates a poisoned copy you cannot modify. The NIST SP 800-209 calls for recording anti-malware scan results for backup copies used for cyber-event recovery. Dedicated pre-backup malware scanning validates data integrity before it enters the isolated environment.
• Treating air gapping as a single-layer defense. Air gapped backups as your sole backup strategy create a single point of failure. Follow the 3-2-1-1-0 rule so no single layer's compromise eliminates all recovery capabilities.

Once you avoid these mistakes, you can move to the operational practices that make the design sustainable.

Air Gapped Backup Best Practices

Choosing the right architecture type gets you started, but operational discipline is what keeps backup isolation effective over time. These eight practices address the areas where air gapped backup designs most often degrade after initial deployment.

1. Classify data before designing architecture. Not all data needs the same air gap approach. Organize backup requirements by regulatory obligation, business criticality, Recovery Point Objective (RPO), and retention schedules.
2. Define RPO/RTO and test against them. If your RPO is one hour, backups must run hourly or faster. If your RTO is four hours, your air gapped restore process must complete within that window, including physical media retrieval. Document these numbers and validate them through scheduled exercises.
3. Implement immutability on all backup repositories. Enable write-once or locked configurations that cannot be modified during retention periods. This complements air gapping by preventing modification even if an attacker gains access to the isolated environment. Used together, immutable backups and air gapped backups give you stronger ransomware recovery options.
4. Enforce MFA and dual-approval for all backup operations. Every access path to backup infrastructure requires MFA, including administrative access. Destructive operations such as deletion, retention policy changes, and immutability disabling should require dual approval from separate administrators. Strong identity security controls are especially important when your backup isolation depends on privileged workflow controls.
5. Scan backups with current tools before and after storage. Periodically re-scan historical backup copies with updated anti-malware tools. This identifies poisoned copies containing malware that was not identifiable at backup time. Behavioral AI from your endpoint protection platform provides a validation layer here.
6. Monitor for backup-targeting behavior. CISA specifically recommends monitoring for anomalous usage of vssadmin.exe, bcdedit.exe, wbadmin.exe, fsutil.exe with deletejournal, and wmic.exe with shadowcopy or shadowstorage commands in its CISA ransomware monitoring guide. Your XDR platform should flag these as high-priority signals.
7. Test restores regularly, simulate disasters annually. Run full restore testing to isolated environments on a regular cadence. Conduct annual disaster simulations that mimic actual data loss events, not just file-level checks. The NIST IR 8576 specifies annual testing for restoring from backup procedures.
8. Restore to isolated staging, not production. Air gapped recovery copies should restore into a dedicated isolated environment where you verify systems are clean before returning them to production. Run behavioral anomaly tools in the staging environment to confirm no reinfection occurs before reconnection.

These practices make your backup design more resilient in real incidents. They also set up the incident examples that show what happens when recovery paths are reachable or unverified.

Real-World Ransomware Incidents and Backup Lessons

Real incidents make the backup risk concrete.

1. Norsk Hydro, 2019, LockerGoga ransomware. When LockerGoga hit in March 2019, Norsk Hydro declared a company crisis and switched to manual operations across 40 countries as ransomware locked files on thousands of servers and PCs. The company later reported NOK 550–650 million losses for the first half of 2019. The incident showed how operational disruption can spread across an entire organization even when production continues in degraded mode — and how backup resilience allowed Norsk Hydro to restore systems without paying the ransom.
2. Colonial Pipeline, 2021, DarkSide ransomware. The company shut down pipeline operations after the attack, and the DOJ later recovered $2.3 million of the 75 Bitcoin ransom payment. CISA's DarkSide advisory documented how the attack on business systems drove major operational disruption across critical infrastructure.
3. MGM Resorts, 2023, social engineering and ransomware-linked disruption. After attackers used social engineering to compromise MGM's identity infrastructure, the company shut down operations across hotel and casino properties to contain the breach. MGM later reported a $100 million EBITDAR impact for September 2023. The event showed how identity compromise can cascade into broad outage conditions, which is why backup isolation alone is not enough without identity controls.

These incidents lead directly to the final question: how do you pair recovery resilience with controls that stop attackers before they can poison, delete, or reach your backup paths?

Key Takeaways

Air gapped backups provide an architectural way to keep at least one recovery copy beyond the reach of routine network-based attacks. Isolation is an essential part of any defensible ransomware recovery plan. 

Implement air gapped backups following the 3-2-1-1-0 rule, test restores regularly, scan before and after storage, and pair backup isolation with Behavioral AI prevention to stop threats before they reach your backup infrastructure. If you rely on offline backups and immutable backups as part of your broader cyber resilience plan, you still need regular validation to support reliable ransomware recovery.

FAQs

What are Air Gapped Backups?

Air gapped backups are isolated copies of critical data that are physically or logically separated from production networks. Physical air gaps use offline media like tape that is removed and stored off-site. Logical air gaps use network segmentation, strict access controls, and immutability to reduce attacker reach. 

Both approaches aim to keep at least one recovery copy beyond the reach of network-based attacks, including ransomware that specifically targets backup infrastructure before triggering production encryption.

What is the Difference Between Immutable Backups and Air Gapped Backups?

Immutable backups prevent modification or deletion after data is written, but they can still remain reachable over the network. Air gapped backups isolate data from routine network access, yet they do not guarantee the copy was clean when captured. 

You need both controls for stronger resilience: immutability helps block tampering, while air gapping reduces attacker reach. Your ransomware recovery plan should also include malware scanning, access controls, and restore testing.

How often should you test air gapped backup restores?

The NIST SP 800-209 recommends regular integrity testing for critical data, and the NIST IR 8576 specifies annual testing for restoring from backup procedures. 

In practice, you should validate critical backup integrity regularly, run full restore exercises in isolated staging at least annually, and measure actual recovery time against your documented RTO and RPO targets during every exercise. More frequent tests make sense for your most critical systems.

Can cloud backups qualify as air gapped?

Cloud backups can qualify as logically air gapped when you isolate them through separate security domains, independent authentication, object-level immutability, and API-only access with MFA. Standard cloud storage alone does not create an air gap. 

You need to configure the separation deliberately, document the control boundaries clearly, and verify that your production credentials cannot directly access, alter, or delete the protected backup set during normal operations.

What is the biggest risk with air gapped backup implementations?

The biggest risk is assuming you have a true air gap when you actually have logical isolation with autonomous workflows. Any scheduled replication job, API call, or backup agent connection creates a reachable pathway that sophisticated attackers may exploit. 

You should audit your architecture honestly, classify it correctly, and add compensating controls such as immutability, MFA, segmentation, approval workflows, and repeated restore testing so your recovery design matches your real exposure.

How do air gapped backups fit into a zero trust architecture?

Air gapped backups align with zero trust principles by enforcing explicit verification at the backup infrastructure layer. Every access request requires authentication, authorization follows least-privilege RBAC, and activity remains logged and auditable. 

The air gap adds a stronger restriction by removing routine network access between backup cycles. Dual-approval workflows for destructive operations add another control, which helps keep a single compromised account from destroying your recovery capability or weakening your cyber resilience posture.