Security improvements should protect the process while respecting uptime, safety and recovery needs.
Operational technology (OT) includes systems that monitor or directly affect physical processes: industrial control systems, PLCs, SCADA, distributed control systems, building automation and related equipment. OT security differs from ordinary enterprise IT because availability, deterministic behaviour, safety and vendor support constraints can be critical. That does not make security optional; it changes how controls are selected, tested and maintained.
Start with an inventory and process context
You cannot manage the exposure of assets you do not know exist. Build an inventory of controllers, engineering workstations, operator stations, servers, network devices, remote access paths and dependencies. Record owner, function, firmware/software version, criticality and support status. A spreadsheet is a valid beginning; an inventory that no one updates is a museum exhibit.
Reduce unnecessary connectivity
Map the legitimate communication flows between zones and conduits, then restrict the rest. Separate control networks from business networks with reviewed boundaries. Remote access should use approved, authenticated and monitored paths, with time-limited access where appropriate. Avoid broad “any-any” firewall rules that become permanent because no one remembers why they were created.
Protect engineering changes
Use individual accounts where feasible, control privileged access, verify backups and test restoration. Store project files and controller backups under version control or another controlled repository. Establish a review and approval process for logic changes, firmware updates and changes to safety-relevant or validated systems. An untracked online change can be operationally useful today and a serious diagnosis problem next month.
Patch based on risk
Not every OT asset can be patched on the same schedule as an office laptop. Check vendor guidance, vulnerability exposure, compensating controls, maintenance windows and rollback options. If a patch cannot be applied promptly, document the risk and the alternative measures—such as segmentation, disabled services, tighter access or heightened monitoring—rather than treating the lack of a patch as the end of the conversation.
Prepare for recovery
Backups should cover engineering projects, recipes or configurations as appropriate, system images, databases and required licenses or recovery information. Test that the backup can be restored to a compatible environment. Define who can authorise shutdown, who contacts the vendor, and what a safe degraded mode looks like. Incident response in OT has to account for the physical process, not only the compromised endpoint.
Use a recognised framework
NIST SP 800-82 Rev. 3 provides guidance on securing OT while accounting for operational requirements; the ISA/IEC 62443 family addresses industrial automation and control system security. Use the current authoritative versions and apply them to the facility’s actual risks and architecture.
Security is a lifecycle activity. A firewall installed during commissioning is not a complete security programme, just as a checked alarm is not a maintenance strategy.