Recovering from ransomware on virtual machines

How to recover from a ransomware attack that targets virtual machines is not a simple matter of restoring the latest disk image. Ransomware may encrypt guest files, alter base templates, abuse hypervisor management tools, steal service-account credentials, or destroy replicated storage. Because one compromised identity can move laterally across dozens of workloads, restoring a single affected VM while production trust remains intact can recreate the breach.

The response must join cyber containment, evidence preservation and service restoration without assuming that a successful reboot equals recovery. This is especially relevant to Australian organisations whose virtual estates span Sydney and Melbourne hosting regions, branch offices and remote operations in resource and logistics sectors. CARM Security, developed by Exclusive Networks, integrates vendor technologies that help teams identify, contain, respond to and remediate attacks across these environments.

Contain the attack without destroying evidence

Treat the event as a major cyber incident whenever encryption, privileged access or shared infrastructure is involved. Appoint an incident commander, a recovery lead and a scribe; establish a single incident channel; and keep a timestamped log of every observation and action. The team should include virtualisation, identity, network, application, legal and communications staff, with the cyber insurer and relevant technology suppliers contacted early. If encryption is still spreading, containment takes priority over perfect forensic collection.

Contain at several layers rather than only disconnecting affected guests. Restrict access to vCenter, ESXi, cloud control planes, storage arrays and administrative interfaces; revoke exposed sessions and API tokens; rotate privileged credentials; and quarantine affected VMs from production networks. If the management plane appears compromised, move recovery to a trusted administrative path instead of reusing it. Suspect templates, disks and snapshots should be suspended or segregated, not deleted, because they can contain evidence and may reveal the attack path.

Preserve evidence from selected systems before they are rebuilt: capture volatile data where feasible, retain relevant logs and create forensic images under documented procedures. Avoid mounting encrypted disks in ordinary workstations. Containment also runs alongside Australian notification duties. Significant cyber security incidents should be reported to the Australian Cyber Security Centre, and suspected criminal activity can be referred to law enforcement. If a notifiable data breach is likely to cause serious harm, OAIC notification must occur as soon as practicable and no later than three calendar days after the threshold is met. APRA-regulated entities must assess and meet applicable notification requirements for material service impacts.

Establish a clean recovery environment

Build the recovery environment in a clean administrative and network domain. This may be a new bare-metal cluster, isolated cloud subscription or separate virtual network, but it must not trust production identities, management certificates or orchestration tools. Patch the hypervisor and management layer, disable unnecessary remote access and give recovery administrators fresh credentials. Maintain an inventory of dependencies such as DNS, directory services, storage, databases, certificate authorities, backup platforms and DevOps pipelines; missing one of these can block a technically successful restore.

Verify backup independence before trusting any copy. Ransomware operators often target backup repositories, snapshots and credentials, so an apparently successful backup job may still point to altered data. Use immutable or offline copies with separate administration and multifactor authentication, identify the last point before attacker access, and test representative restores. Record whether restore points include guest data only or also application-consistent database states. A restore performed into the original production environment should be treated as suspect, not as proof of cleanliness.

Rebuild rather than recycle whenever templates, guest operating systems or identities may have been altered. Start with known-good images or infrastructure as code, then apply current security baselines and approved application packages. Keep the original systems available for comparison through controlled evidence procedures. Within a CARM Security-led response, teams can draw on multiple vendor technologies to reduce visibility gaps across affected workloads. Exclusive Networks’ partner ecosystem also reflects the Australian market’s reliance on specialist distribution, managed-service support and collaboration between internal security teams and external specialists.

Choose the safest restoration path

The recovery decision depends on backup integrity, encryption coverage, business downtime tolerance and whether a trustworthy clean-room environment is available. Comparing the practical routes makes trade-offs explicit before production data is moved.

Recovery route When it fits Essential controls Principal limitation
Restore known-good backups Immutable copies exist from before unauthorised access Isolated restore, malware scanning, patching, credential replacement and application testing Residual malware or stolen credentials may survive
Rebuild from trusted images or infrastructure as code Standard workloads have maintained golden images or automation Fresh instances, current baselines, secret rotation and configuration comparison Hidden drift and incompatible dependencies can delay service recovery
Use a clean replacement environment Extended outages or fully untrusted originals make direct restoration unsafe New network, separate administration, independent data recovery and controlled migration Operational divergence and reconciliation become more complex

Select the route using evidence, not optimism. If a known-good backup and clean environment are available, restore into isolation, scan and test the workload, then migrate it in stages. If credentials, templates or management infrastructure were stolen, restoring guest disks alone is insufficient: rotate application secrets, SSH keys, tokens, certificates and database credentials before any workload regains connectivity.

Where clean copies do not exist, rebuild critical workloads and recover data from independent sources such as object storage, databases with external replication, mail systems or supplier-provided exports. Treat any attacker-supplied decryption tool as untrusted until its behaviour has been examined, and remember that decryption does not remove persistence mechanisms or stolen data. Run forensic analysis and service restoration in parallel so evidence collection does not become an indefinite delay, while insurer-approved specialists support decisions under the incident policy.

Validate recovery before reconnecting services

Define release criteria before reconnecting restored systems. Each VM should pass malware scanning, configuration comparison, patching, logging and application-owner validation, while recovered accounts receive new credentials and newly issued tokens. Introduce restored workloads through a staged network path, beginning with test users and low-risk services. Map each application to its supporting VMs and shared services so teams do not restore a front end while leaving its identity provider, database or storage dependency exposed.

Assess possible exfiltration as well as encryption. Review hypervisor, cloud, identity, DNS, firewall and endpoint logs for unusual archive creation, bulk downloads, new cloud storage destinations and privileged logins. Preserve evidence of suspicious access and engage specialist responders when the scope is unclear. A clean disk can still sit within a compromised identity architecture, which is why access paths and trust relationships need the same scrutiny as the files themselves.

For organisations in Australia, the business impact can extend beyond office systems. A hospital in Brisbane may need clinical services to remain available through the night, while a Pilbara mining operator in Western Australia may depend on remote connectivity and local safety procedures. These environments require prioritised recovery, tested manual workarounds and coordination with sector obligations. Once service continuity is established, map the estate against the Australian Signals Directorate’s Essential Eight rather than treating incident recovery as a one-off technical project.

Make the recovery repeatable

Use the incident to correct the conditions that enabled entry and spread. The post-incident review should identify the initial access route, affected administrative identities, control-plane weaknesses, backup exposure and gaps in logging. Then implement priority changes such as removing unnecessary internet exposure, enforcing phishing-resistant multifactor authentication for privileged access, isolating management networks, improving immutable backup coverage and deploying EDR with tested response playbooks. CARM Security supports post-breach remediation and mitigation by bringing vendor capabilities into a coordinated ecosystem.

Rehearse the decisions as well as the technology. Update the business continuity and disaster recovery plan with VM dependencies, recovery order, minimum viable services, supplier contacts and evidence-retention responsibilities. Run a scenario in which attackers hold the virtualisation console and corrupt backup repositories across a distributed Australian team, including handoffs between Sydney, Brisbane and Perth. CARM Security documentation, videos and demonstration sessions can support technical workshops and tabletop exercises for internal teams, service providers and channel partners.

Completion should mean an accountable owner and deadline for every corrective action, not simply that the last production VM is switched on. The next concrete step is to open the incident bridge, appoint an incident commander and recovery lead, and record the first three containment decisions—network isolation, privileged credential rotation and backup verification—with a named owner and deadline for each.