OT cyber security has moved beyond trying to prevent incidents as prevention is impossible to guarantee. With industrial and manufacturing businesses increasingly becoming targets, the focus is now on cyber resilience and knowing what to do once defences have been breached.
Now, prevention remains important, but systems are becoming more interconnected and remote access is added to legacy technology, so the threat landscape is changing. This shift is reflected in the latest guidance from the National Cyber Security Centre (NCSC). For OT specifically, the NCSC warns that recovery can be slower and more complex than IT. This is because measures must consider safety, operational constraints and system dependences that will affect the order that systems can be restored.
Know the blast radius
Knowing that an asset has been compromised is only useful if the business understands what assets it connects to and how it could affect them. Looking at the example of a typical HMI, businesses must ask important questions:
- Which PLCs does it communicate with?
- Does it exchange information with a historian or supervisory system?
- Can it communicate with systems elsewhere in the facility?
- What other assets depend on the information it provides?
The answers determine whether an incident can be contained around one asset or has the potential to spread. To prepare for cyber resilience, businesses should go beyond a basic inventory of OT assets by mapping the communication pathways and dependencies, creating a complete asset inventory. With that in place, they can establish a blast radius of an incident.
Mapping communications regularly is important as new connections are added between OT and IT and any connection that provides an operational function can also provide cyber attackers with a route through the system.
Mapping relationships before an incident also means that when suspicious activity is detected security and operations teams are not battling on two fronts: understanding the architecture while simultaneously responding to an attack. Businesses can quickly identify which other systems could be affected and the operational consequences of isolating an asset. Obviously removing every connection from a connected environment is not possible. It is possible however to make every connection understood. During an incident, this will help create boundaries to stop an incident from spreading.
Grouped functions
Not every asset in an industrial environment needs to communicate with every other, even if they are in the same network. Systems can be grouped according to their function, criticality, and security requirements. This approach is known as network segmentation. IEC 62443-3-2 provides a framework for dividing control systems into zones and conduits based on risk.
For businesses already working to IEC 62443 principles with defined zones, it’s important to ask whether those zones still represent the plant today. Facilities are constantly evolving as equipment is added or replaced, processes change, new suppliers are granted access and OT systems are connected to wider business systems. Over time, new connections accumulate and although they may have a legitimate purpose, they can influence how far an incident can reach. This makes containment something that needs to be validated instead of assumed.
OT professionals should regularly review their control zones and test whether the original segmentation is still fit for purpose by asking a series of questions. If suspicious activity was detected in one zone, can communication between conduits be restricted? Which operational functions would be disrupted by isolating that zone? What new dependencies have been introduced since the system was originally designed or last mapped? Could an attacker use an authorised path to reach another zone?
Segmentation does not necessarily prevent the initial compromise. Its value comes from preventing that one compromise from becoming a site wide problem. Businesses should regularly test their architecture against worst case scenarios and map dependencies between systems. This can identify where boundaries will limit an attacker’s movement. This is particularly important for brownfield sites where it is unrealistic to replace every legacy asset with connected technology, so it’s important to control what can communicate with those assets, limiting the impact of a compromise.
Businesses must also consider whether a cyber attacker might find a route in by compromising a legitimate identity. An engineer may require access for a specific reason but if those credentials are compromised then attackers could further their reach. Privileged access must be considered regularly as part of defining the blast radius, especially if the same identities can be used across IT and OT environments.
Can you operate while part of the plant is unavailable?
Successfully containing an incident however does create a new challenge. If part of the OT environment has been isolated, the team needs to ensure that the rest of the operation continue without it.
Disconnecting an affected system or restricting communication between zones can remove data, monitoring, or control that another process relies on. This is why the operational consequences of segmentation need to be understood before an incident takes place. Businesses must establish which functions are critical, which can be temporarily unavailable and whether processes can continue in a degraded state.
If a supervisory system becomes unavailable for example, can processes continue under local control? If central monitoring is lost, can teams maintain visibility? Where manual intervention is needed, has it been tested and do the people responsible know how to respond?
Minimum viable operations
The NCSC describes the objective during recovery as ‘reaching minimum viable operations’. Reaching that point means understanding the minimum technology, people and information needed to support critical functions.
Rather than only rehearsing the technical response to an attack, businesses should simulate the loss of individual systems or zones. One process could potentially operate locally for hours without a supervisory system, while another may need to be brought to a safe stop immediately. Knowing the difference gives the response team a clear basis for what needs to be contained, what can continue operating and what should be prioritised.
Authority also needs to be set as part of this plan. If an affected production area needs to be isolated at 2am, who is authorised to make that decision? Who determines what connected processes can continue safely or if an incident requires a wider shutdown?
These decisions are not solely up to cyber security teams, they require input from the people who understand the process and its safety implications. An incident response plan that says an affected system should be isolated will not protect the operation if nobody has established who is responsible for taking action. The goal should therefore be to restore the right functions, in the right order and to minimum viable operations.
For more articles like this, visit our Digital Transformation channel



