Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
The Purdue model describes an industrial enterprise as six functional levels, from field instruments to corporate IT, and it remains the common language for segmenting control networks.

What Is the Purdue Model?
The Purdue model is a layered description of an industrial enterprise, running from the valves and sensors touching the process up to the corporate systems that plan production. Theodore J. Williams and his colleagues at Purdue University published the Purdue Enterprise Reference Architecture in 1989 as a manufacturing integration framework, and it later fed into ISA-95. It was never written as a security model. It became one because it turned out to be the most convenient way to draw a line between things that must not talk to each other.
| Level | What lives there | Typical security concern |
|---|---|---|
| Level 0 | Sensors, actuators, valves, motors, drives | Physical access, fieldbus with no authentication |
| Level 1 | PLCs, RTUs, DCS controllers, safety-instrumented systems | Unauthenticated programming, fragile network stacks |
| Level 2 | HMIs, SCADA servers, engineering workstations, local historians | Windows estate, removable media, vendor toolchains |
| Level 3 | Site operations: manufacturing execution, batch management, plant historian, domain services | The usual IT risks with OT consequences |
| Level 3.5 | Industrial demilitarised zone | The single most important boundary in the architecture |
| Level 4 | Business systems, ERP, email, file services | Where almost every industrial incident begins |
| Level 5 | Enterprise and external connectivity, cloud services | Internet exposure and third-party integrations |
Levels 0 to 2 are the cell or area. Level 3 is the site. Levels 4 and 5 are the business. A safety-instrumented system is sometimes drawn as a separate Level 1 branch, and in a well-designed plant it has its own network entirely.
The demilitarised zone between Level 3 and Level 4 exists so that no session ever crosses directly between business and control networks. Data that the business needs — production figures, historian replication, alarm summaries — is published into a broker in the DMZ, and business systems read it from there. Control systems never originate outbound connections to the enterprise, and enterprise systems never reach inwards.
When this boundary is drawn correctly, ransomware in the corporate estate is contained to the corporate estate. When it is drawn on a diagram but not in the switch configuration, it is not a boundary. Architecture reviews routinely find a domain controller serving both sides, a backup agent reaching down into Level 2, or a remote access tool bridging the whole stack, any one of which makes the DMZ decorative.
The model is a vocabulary, not a blueprint. Its practical value is that it gives engineers and security teams a shared way to say where something sits and therefore what it is allowed to do. In IEC 62443 terms, Purdue levels inform but do not replace zones and conduits: two pieces of equipment at the same level may belong in different zones if their consequence of failure differs, and the formal partitioning exercise in IEC 62443-3-2 is what produces an enforceable design. Use Purdue to orient, use zones to build, and use network segmentation to enforce.
Several organisations have declared the model obsolete on these grounds. In practice it has survived because nothing has replaced its communicative value, and the sensible position is the one taken by most industrial defenders: keep Purdue as the shared map, add explicit zones for the connections that break the hierarchy, and make sure every one of those exceptions is deliberate, inspected, and recorded in the asset inventory.
UK operators of essential services will find the NCSC’s Cyber Assessment Framework asks for exactly this reasoning under Principle B4: not whether you follow a particular architecture, but whether the separation you claim is designed, enforced, and verified. At Privia Security, our Infrastructure and Architecture Review Services test that claim against the running configuration.
As a shared vocabulary, yes. As a strict hierarchy, less so, because cloud telemetry, industrial IoT, and remote operations create legitimate connections that skip levels. Most practitioners keep the model for communication and layer formal zone and conduit definitions on top, treating every level-skipping path as an exception that must be justified and controlled.
Broker and replication services that let the two sides exchange data without a direct session: a historian mirror, a patch and antivirus distribution point, a remote access jump host, a file transfer service, and an update repository. Nothing in the DMZ should be essential to running the process if it fails.
They often do on a shared virtualisation platform, which is acceptable if the hypervisor and its management network are treated as the most sensitive component in the design. The risk is a management interface reachable from the business network, which collapses the separation entirely regardless of how the virtual networks are configured.
Purdue describes function; IEC 62443 describes security requirements. The standard asks you to partition the system under consideration into zones that group assets with common security needs and conduits that carry traffic between them. Purdue levels are a good starting point for that partitioning, but consequence of failure, not level, decides the final boundaries.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.