OT Pentest Framework

The Framework

How the OT Penetration Testing Framework works

IT penetration testing has well-known methodologies. OT needs its own, because the targets aren't only data. They're pumps, breakers, valves, furnaces and the people working beside them.

Why a separate framework for OT?

The framework took inspiration from its IT cousin: learn the target, get in, find what matters, and prove the risk. But in OT, the end goal is usually not stealing data. It's influencing a physical process. That changes what attackers go after, how carefully testing has to be done, and what a "finding" really means.

An OT network still contains plenty of traditional IT, especially Windows-based systems. Alongside them are OT/ICS assets such as programmable logic controllers (PLCs), human-machine interfaces (HMIs), distributed control systems (DCS), safety instrumented systems (SIS) and supervisory control and data acquisition (SCADA) systems, talking protocols like Modbus, S7, DNP3, EtherNet/IP and BACnet. Many of those protocols have no authentication at all.

Mapping the environment: the Purdue Model

The Purdue Enterprise Reference Architecture (PERA) was originally written to describe timing requirements between assets in OT environments compared with IT networks. Over time it was adapted for cybersecurity discussions, and that "expanded" Purdue Model is how the framework describes where an attacker is.

  1. 5LevelEnterprise networkCorporate IT, email, ERP, cloud and internet connectivity
  2. 4LevelBusiness & site ITSite business systems, file shares, Active Directory, user workstations
  3. 3.5LevelIT/OT DMZJump hosts, patch and antivirus relays, historian replicas, remote access gateways
  4. 3LevelSite operationsEngineering workstations, historians, OT servers, OT domain controllers
  5. 2LevelSupervisory controlHMIs, SCADA servers, alarm and operator stations
  6. 1LevelBasic controlPLCs, RTUs, DCS controllers, safety instrumented systems
  7. 0LevelPhysical processSensors, actuators, motors, pumps, valves and breakers
The expanded Purdue Model. Attackers usually start in IT at level 4 or 5 and work down through the DMZ toward the controllers. Phase 1 attacks cover levels 5 to 3. Phase 2 attacks touch levels 2 to 0, where the physical process lives.

Phase 1 and Phase 2 attacks

Phase 1: get into position

  • Main goals are to gain access to control systems and gather information about the industrial process
  • Mostly involves traditional IT assets at levels 3 to 5
  • Puts the attacker in a position to attack OT assets at levels 0 to 2
  • Largely follows the IT penetration testing methodology

Phase 2: touch the process

  • Direct interaction with the OT assets that monitor and control a process
  • Requires technical and engineering knowledge to be effective
  • Builds on information gathered during Phase 1
  • Can lead to high-impact incidents

The "assumed breach" scenario

Most real-world OT penetration tests start with an assumed breach: the tester is given access to the IT network, on the reasonable assumption that a determined attacker will get there eventually, whether through Wi-Fi, an insider, a physical breach, phishing, an exploited vulnerability, other social engineering or compromised remote access credentials.

From there, a test may include getting onto a system in the IT/OT DMZ, establishing a foothold on the OT network at level 3, and moving laterally through the higher levels. Testing almost always stops, at the latest, once the tester is in a position to interact with control systems: HMI access, a SCADA interface, or the ability to talk to a PLC. Phase 2 techniques belong in a lab.

The realities of OT penetration testing

  • Demand for OT/ICS penetration testing is limited, and most operations can't afford one.
  • Most environments need a risk assessment or a maturity assessment, not a pentest.
  • Most "OT penetration tests" are really IT penetration tests that end at the OT boundary.

That's exactly why the framework is aimed at defenders. Knowing the full attack chain, including the Phase 2 steps no one should test in production, tells you where to put your effort.

Using the framework with MITRE ATT&CK for ICS

The phases line up with the tactics in MITRE ATT&CK for ICS, with Reconnaissance added from the IT testing world. Use the framework to walk the attack chain in order, and ATT&CK for ICS to dig into the specific techniques and real-world examples behind each phase.

Safety and authorization come first

The information on this site is for educational purposes only. Any security testing must be performed with authorization and in close coordination with the right engineering and operations team members. Testing should never be performed in a production environment where safety or operational availability could be affected.

Walk the 11 phases