Isolating Critical Assets During a Targeted APT Campaign
A targeted advanced persistent threat (APT) campaign rarely announces itself with a single dramatic alert. Attackers may spend weeks harvesting credentials, mapping identity systems, testing remote access and quietly moving towards the systems that matter most. By the time a security team sees unusual PowerShell, suspicious authentication or a new administrator account, the compromise may already involve several business units.
Asset isolation is the disciplined separation of high-value systems from the attacker’s paths. It can involve network segmentation, identity restrictions, endpoint containment, application controls and temporary changes to operational processes. The aim is to limit blast radius while preserving enough business function for essential services to continue.
For Australian organisations, this work sits within a practical regulatory environment. The Australian Signals Directorate’s Essential Eight encourages controls such as application control, restricted administrative privileges and regular patching. APRA-regulated entities must consider CPS 234 obligations, while organisations covered by the Security of Critical Infrastructure Act may face additional risk-management and incident-reporting requirements. A response plan needs to connect technical action with those obligations.
Identify The Assets That Matter Most
Isolation begins with a business view of criticality rather than a list of every device in a configuration management database. An identity provider, privileged access management platform, backup console, domain controller, payment system or clinical application may create a much larger risk than an ordinary workstation. A compromise of one control plane can give an intruder reach across the environment.
Classify assets by the harm caused if they are altered, encrypted, exposed or made unavailable. Useful categories include safety-critical systems, revenue-generating applications, sensitive data stores, identity infrastructure, management tools and recovery services. Include dependencies: a warehouse application may rely on Active Directory, DNS, a database cluster and a third-party remote support service.
Australian operating models make dependency mapping especially important. A mining company in Western Australia may have a remote operational site linked to Perth and Brisbane, while a regional hospital may depend on shared health services and vendors outside its immediate network. A council or utility may use a managed service provider for security operations. These arrangements can make a seemingly simple disconnection disruptive if ownership and fallback procedures are unclear.
A practical asset register should record the business owner, technical owner, data classification, normal communication paths, backup location, recovery priority and isolation method. It should also show which systems must remain reachable for safety, public services or legal reporting. This information gives responders permission to act quickly without spending the first hours of an incident arguing about what can be disconnected.
Contain The Attacker Without Destroying Evidence
During a live APT response, containment should be proportionate and staged. An infected laptop can often be removed from the network immediately. A production database, identity service or industrial control system may require a controlled isolation window, a read-only mode or a carefully filtered route. Cutting every connection at once can interrupt operations and erase useful evidence.
Start with the attacker’s observed paths. Disable compromised accounts, revoke active sessions, rotate exposed credentials and block malicious infrastructure at email, DNS, proxy and firewall layers. Apply endpoint network containment to confirmed or high-confidence compromised hosts. For sensitive server groups, use allow-listed communication rather than broad internal access, then monitor every permitted connection.
Evidence handling needs to run alongside containment. Preserve relevant logs from endpoint detection, identity, VPN, cloud, firewall and email platforms before retention periods remove them. Record who authorised each action and when it occurred. Memory capture or forensic imaging may be appropriate for selected hosts, although a safety-critical system should not be destabilised merely to collect a perfect image.
A coordinated platform such as CARM can help bring vendor telemetry and response actions into one operational picture. Its value during post-breach remediation is the connection between identification, containment and recovery: analysts can correlate signals, prioritise affected assets and coordinate controls across technologies. The platform does not replace incident leadership, but it can reduce the delay between finding a compromised path and closing it.
Build Segmentation Around Trust And Function
Segmentation is more effective when it follows business function and trust level. Separate user networks from server networks, administrative workstations from ordinary endpoints, backup infrastructure from production systems, and operational technology from corporate IT. Apply the same principle in cloud environments through separate accounts, subscriptions, security groups and private endpoints.
Identity is central to this design. Network location alone cannot be trusted when attackers possess valid credentials or have established remote access. Require phishing-resistant multifactor authentication for privileged users where possible, use separate administrative accounts, restrict just-in-time elevation and remove standing access to critical assets. Service accounts should have narrowly defined permissions, managed secrets and clear ownership.
Microsegmentation can be introduced in stages. First, identify normal east-west traffic and the services each asset genuinely needs. Next, enforce controls in alert or simulation mode, investigate unexpected dependencies and then move to blocking. For a targeted campaign, create an emergency policy that permits only known management, backup, monitoring and application flows. Review that policy daily until the incident is contained.
Australian organisations often need to balance central security standards with local autonomy. A national retailer may have stores across New South Wales, Victoria and Queensland, each with different connectivity and support arrangements. A temporary isolation design should allow local teams to keep essential operations running while restricting routes back to central identity, finance or administration systems. Clear escalation contacts matter when the security team is in Sydney and a site engineer is working an arvo shift in Townsville.
Protect Recovery Systems And Third Parties
Attackers frequently target recovery mechanisms because they can turn a contained intrusion into a prolonged outage. Backup repositories, virtualisation managers, software deployment tools and privileged administration platforms deserve isolation equal to production systems. Maintain separate administrative identities, restrict backup deletion rights and keep at least one recovery copy disconnected or otherwise protected from routine domain compromise.
Validate that backups are usable before an incident. A successful backup job does not prove that applications can be restored, credentials are available or data is complete. Test representative systems, including identity, databases and customer-facing services. Store recovery procedures where an attacker controlling the corporate environment cannot alter them, and make sure the procedures work when normal email and collaboration tools are unavailable.
Third parties require the same scrutiny. Remote monitoring and management tools, outsourced help desks, cloud administrators and software suppliers may provide legitimate routes into critical assets. Suspend or narrow those routes during an APT campaign, require approval for privileged sessions and review vendor-issued indicators of compromise. A supplier’s access should be restored only after its account, endpoint and authentication path have been assessed.
This is particularly relevant in Australia’s market, where smaller organisations commonly rely on a managed security service provider or a national systems integrator for specialist skills. Outsourcing monitoring does not outsource accountability. Contracts should define notification times, evidence handling, privileged access, data location and authority to isolate systems. For regulated organisations, those terms also support governance and audit expectations.
Turn Isolation Into A Managed Response
A response playbook should define triggers, authority and technical actions before an incident occurs. Triggers may include confirmed credential theft, lateral movement towards domain controllers, unauthorised access to backups, exploitation of an internet-facing service or evidence that a privileged account has been used from an abnormal location. Each trigger should map to a containment tier and an executive contact.
The first tier might restrict a single endpoint or account. The second could isolate a server group, vendor connection or administrative segment. The highest tier may separate an entire business function from corporate networks, with continuity measures activated by the relevant operational owner. Every tier needs an expiry or review time so emergency restrictions do not become invisible permanent architecture.
Measure whether isolation is working. Useful indicators include the time from detection to account revocation, the number of critical assets with unrestricted east-west access, the percentage of privileged sessions using strong authentication, and the time required to restore a clean service. Track blocked connections and attempted access after containment; repeated attempts can reveal another compromised credential or overlooked route.
Run exercises using Australian operating conditions rather than a generic data-centre scenario. Include a public holiday, limited staff in a regional office, a third-party provider in another time zone and a requirement to notify relevant authorities. Test communications through an alternate channel and rehearse how leaders decide whether to keep a service online, place it in a degraded mode or isolate it entirely. A calm, pre-agreed decision process is worth more than an impressive diagram that no one can use at 2 am.
| Isolation approach | Best use during an APT campaign | Main benefit | Primary risk | Control needed |
|---|---|---|---|---|
| Endpoint network containment | Confirmed compromise on a workstation or laptop | Fast reduction of attacker access | The host may be needed for forensic evidence | Preserve logs and document the action |
| Account and session isolation | Stolen credentials, suspicious privileged access or token abuse | Stops identity-led lateral movement | Legitimate users or services may be interrupted | Revoke sessions, rotate secrets and use separate admin accounts |
| VLAN or subnet quarantine | Groups of affected servers or user devices | Limits movement across a defined zone | Attackers may already have routes through shared services | Use allow-listed flows and validate dependencies |
| Microsegmentation or workload policy | Cloud, server and application environments | Precise control over east-west traffic | Incorrect rules can break production | Begin with visibility, then enforce tested policies |
| Third-party access suspension | Compromised supplier route or untrusted remote support | Removes an external attack path | Delays support or essential maintenance | Define emergency contacts and approved fallback access |
| Critical-service isolation | High-impact identity, payment, health or operational systems | Protects the most consequential assets | Business disruption and safety concerns | Use executive authority, continuity plans and recovery testing |
Isolation is successful when it buys the organisation time without creating a second crisis. Critical assets should be identified before the campaign, their dependencies documented, their access paths narrowed and their recovery options tested. During the event, every technical action should have an owner, a business rationale and an evidence trail.
For an Australian security team, the immediate next step is to select one high-value system, map its users and dependencies, define an emergency isolation rule, and test that rule in a controlled maintenance window.