OT vs IT Security: Key Differences for ICS Teams
How OT security differs from IT security: safety and availability first, deterministic timing, 20 year lifecycles, fragile devices, and what to do about it.
Operational technology (OT) is the hardware and software that monitors and controls physical processes: programmable logic controllers (PLCs), remote terminal units (RTUs), distributed control systems (DCS), SCADA servers, human machine interfaces (HMIs), safety instrumented systems (SIS) and the networks between them. Industrial control systems (ICS) are the largest family of OT, alongside building automation, physical access control and transportation systems.
Most security teams arrive in OT with an IT background. The tools look familiar (Windows servers, Ethernet switches, firewalls) but the priorities, constraints and failure modes are different. Applying IT habits without adjustment is one of the most common ways security work causes an outage. This guide covers the differences that matter in practice and how to translate them into decisions.
Safety and availability come first
The classic IT security triad is confidentiality, integrity and availability, often read in that order. In OT the order is effectively reversed, and safety sits above all three.
NIST SP 800-82 Rev. 3 states it plainly in its comparison of OT and IT: for OT the primary concerns include safety and fault tolerance to prevent loss of life or endangerment of public health, and any security measure that impairs safety is unacceptable. A leaked recipe file is a business problem. A controller that stops responding during a batch reaction, or a safety system that is bypassed, can injure people, damage equipment that takes months to replace, or release material into the environment.
In practice this means:
- Integrity of control data matters more than its secrecy. An attacker who changes a setpoint or spoofs a sensor value is far more dangerous than one who reads it. Detection and prevention of unauthorized writes is a higher priority than encrypting telemetry.
- Availability is measured in process terms. "The HMI was down for ten minutes" may be harmless on a slow thermal process and critical on a fast compressor.
- Fail-safe behavior is engineered, not assumed. Before deploying a control that can block traffic, ask what the process does when that traffic stops.
Timing, determinism and fragile stacks
IT networks optimize for throughput and tolerate latency and jitter. Many OT networks carry cyclic I/O traffic where a controller expects a response within a fixed interval and declares a fault if it does not arrive. Industrial Ethernet protocols such as EtherNet/IP implicit messaging (UDP 2222) or PROFINET real-time traffic run on short cycle times, and a switch that adds jitter or a firewall that queues packets can trip a connection timeout.
Embedded devices also have small network stacks and little CPU headroom. Behavior that is routine on IT systems can disrupt them:
- Active scanning with aggressive timing or unusual packets has crashed PLCs and network modules. If you must scan actively, use protocol-aware tools at low rates during an outage window, and test first on a spare. Prefer passive discovery, covered in OT asset inventory with passive network monitoring.
- Broadcast storms from a misconfigured switch can saturate controllers that process every broadcast in software.
- Endpoint agents can consume CPU or lock files that an HMI or historian depends on. Vendors usually publish an approved antivirus list and exclusion settings for their software; follow it.
Lifecycles measured in decades
A laptop is replaced every three to five years. A PLC, a DCS controller or a protection relay is commonly expected to run for 15 to 20 years or more, and some stay much longer because replacing them means re-engineering, re-commissioning and sometimes re-certifying the process. The consequences:
- Unsupported operating systems are normal. Engineering workstations and HMIs often run Windows versions that are out of support because the control software is only validated on them.
- Patches need vendor qualification. The automation vendor often has to test an OS patch with their software before you deploy it. Firmware updates for controllers may require a process shutdown.
- Insecure-by-design protocols persist. Modbus, DNP3 without Secure Authentication, and older S7 and CIP traffic carry no authentication. You will defend them for years with network controls, as explained in ICS protocol security.
This is why OT vulnerability management relies heavily on compensating controls and risk-based prioritization rather than "patch everything within 30 days". See OT patch and vulnerability management.
Change management is a safety process
In IT, a configuration change can often be rolled back in minutes. In OT, a change to a controller program, a network path or a firewall rule can alter how the process behaves. Mature sites route changes through a management of change (MOC) process involving operations, process safety and engineering. Security changes belong in that process too.
Practical habits that earn trust with operations:
- Propose changes with a clear rollback plan and a defined test, for example "verify that HMI tags update and that the historian keeps collecting for 15 minutes".
- Schedule disruptive work in planned outages or turnarounds, which may be months apart.
- Deploy new controls in monitor or alert mode first (firewalls in logging mode, intrusion detection passive) before enforcing.
- Keep the engineering team in the loop. They know which "odd" traffic is a 20 year old integration that still feeds a regulatory report.
Different people own the systems
Control networks are typically run by control or automation engineers, not by IT. They answer to plant management and are measured on uptime, quality and safety. Security teams that show up with policies and scanners and no process knowledge get ignored or blocked.
The organizations that succeed form a cross-functional team: IT security, OT engineering, operations, process safety and procurement. NIST SP 800-82 Rev. 3 recommends this cross-functional approach when establishing an OT cybersecurity program, and IEC 62443-2-1 frames the asset owner's security program the same way. See IEC 62443 explained and the NIST SP 800-82 overview.
Physical consequences change the threat model
In IT, the worst outcome is usually data loss, fraud or downtime. In OT, adversaries can aim for physical effects: tripping breakers, damaging rotating equipment, or disabling the safety layer that protects people. MITRE ATT&CK for ICS captures this with tactics that have no IT equivalent, such as Inhibit Response Function and Impair Process Control, and impacts such as Loss of Safety and Damage to Property. See MITRE ATT&CK for ICS.
The real incidents covered in ICS attack case studies illustrate this: Stuxnet manipulated centrifuge drive frequencies while showing normal values to operators, the 2015 Ukraine attack opened breakers through the operators' own SCADA tools, and TRITON targeted safety controllers directly.
That said, most OT disruptions observed in practice still start in IT. Ransomware on the business network can force a plant to stop production because shared services (Active Directory, ERP, file shares, remote access) are compromised or because the organization shuts down OT out of caution. The IT and OT boundary is therefore the most important place to invest. See OT network segmentation and the industrial DMZ.
Side by side
| Topic | Typical IT | Typical OT |
|---|---|---|
| Top priority | Confidentiality and integrity of data | Safety, then availability and integrity of control |
| Performance | High throughput, tolerant of delay | Deterministic timing, low jitter |
| Asset lifetime | 3 to 5 years | 15 to 20 years or more |
| Patching | Regular, often automated | Vendor-qualified, scheduled in outages |
| Rebooting to fix | Normal | Often unacceptable during operation |
| Protocols | Modern, authenticated, encrypted | Many legacy protocols with no authentication |
| Testing in production | Common | Avoided; use test benches or spares |
| Owners | IT department | Engineering and operations |
| Worst case | Data breach, downtime | Injury, environmental release, equipment damage |
What carries over from IT
Not everything is different. Many IT controls work well in OT when adapted:
- Asset inventory is the foundation in both worlds. In OT it is built mostly from passive network data and engineering documentation.
- Network segmentation and firewalls are even more valuable in OT because endpoints cannot defend themselves.
- Strong authentication for remote access, with multi-factor authentication and session recording, applies directly. See secure remote access for OT vendors.
- Application allowlisting fits OT servers and HMIs well because their software set rarely changes. CISA recommended it after both the 2015 Ukraine and CrashOverride incidents.
- Backups of controller programs, configurations and golden images are essential for recovery. Test restores, not just backups.
- Logging and monitoring work, but rely more on network sensors than on endpoint agents.
- Incident response needs an OT-specific plan with operations involved and safety-first decision points. See OT incident response.
A practical first 90 days
If you are starting OT security at a site, this order tends to work:
- Learn the process. Walk the plant with an operator. Understand what is produced, what can go wrong physically and which systems the operators cannot work without.
- Map the network and assets from existing drawings, switch configurations and passive traffic capture. Expect the documentation to be out of date.
- Find the IT and OT connections: every firewall, dual-homed host, remote access path, cellular modem and vendor connection. These are where attackers enter.
- Close or control remote access: remove persistent vendor tunnels, enforce MFA, route sessions through a jump host in the industrial DMZ.
- Back up controller logic and configurations and store them offline.
- Write a first OT incident playbook with operations, covering ransomware in IT and suspicious activity in OT.
Then move on to a structured framework: the Purdue model for reference architecture, IEC 62443 for zones, conduits and security levels, and NIST SP 800-82 for program structure.
References
- NIST SP 800-82 Rev. 3, Guide to Operational Technology (OT) Security
- CISA: Industrial Control Systems
- CISA: ICS Recommended Practices
- ISA/IEC 62443 Series of Standards (ISA)
- MITRE ATT&CK for ICS matrix
- SANS: The Five ICS Cybersecurity Critical Controls
- CISA alert IR-ALERT-H-16-056-01: Cyber-Attack Against Ukrainian Critical Infrastructure