How to Secure OT and IT Networks on the Factory Floor
Step-by-step guidance on segmenting OT and IT networks in production facilities to stop malware from spreading from office computers to plant machinery.
Why the factory floor is now an IT problem
For years the shop floor ran on its own island. Programmable logic controllers, CNC machines, HMIs, and sensors spoke proprietary protocols to each other and to almost nobody else. The office network lived in a separate building, on a separate switch, on a separate budget.
That separation is gone. Manufacturers now push production data into ERP and inventory systems, run quality reporting through the same cloud tools the front office uses, and let engineers check a line from a laptop or a phone. Every one of those connections is useful. Every one of them is also a path that malware can travel.
The result is a risk that most small and midsize manufacturers underestimate: an infected office workstation is no longer just an office problem. Ransomware that lands on an accounts-payable PC can crawl through the shared link into the control network, and machinery that runs a twenty-year-old operating system does not patch on demand. It simply stops.
This guide walks through how to secure that boundary in practical steps - the kind of work done during a normal production schedule, not a multi-month shutdown.
Understand what you are actually protecting
Operational technology (OT) and information technology (IT) are built on opposite priorities.
IT systems prize confidentiality and are patched constantly. OT systems prize availability and uptime - a controller that reboots mid-run can scrap a batch or damage equipment. That difference is why you cannot simply apply office security policy to a production line, and why the goal is not to wall the floor off entirely. The goal is to control, log, and limit every connection between the two zones.
Two things drive the plan: what breaks if a system goes down, and who or what genuinely needs to talk to it.
Step 1: Map your systems and dependencies
You cannot segment what you have not documented. Start with a simple inventory:
- Every device on the production network - controllers, drives, HMIs, scale heads, label printers, cameras, and the industrial switches that connect them.
- Every device on the corporate network - servers, workstations, phones, access points, and any personal devices staff have attached.
- Every link between the two - the crossover cable into the ERP server, the engineer's remote-access tool, the shared file share, the vendor's support tunnel.
For each link, write down who uses it and what business process it serves. You will usually find that most OT-to-IT traffic is a handful of necessary flows surrounded by convenience connections nobody remembers creating. Those convenience paths are where incidents start.
This map is also the artifact you hand to any IT partner or auditor. If you work with an outside provider, expect them to build it with you rather than guess.
Step 2: Segment the network
Segmentation means placing your OT assets on their own network segment, separate from office traffic, and forcing all communication between them through a controlled point.
In practice that means separate VLANs for production, corporate, guest, and management traffic, with routing that only permits the flows you approved in step one. A modern industrial or business-grade firewall sits between the zones and decides what crosses.
A few rules that keep this practical:
- Default deny. Start from "nothing crosses" and open only the specific ports and destinations a process requires.
- One-way where possible. Production data should flow to the business systems on a schedule; business systems should rarely initiate sessions back into control equipment.
- Isolate legacy. Machines that cannot be patched belong on the most restricted segment, not on a flat network with everything else.
- Separate guest and personal devices. Your guest Wi-Fi should never share a segment with either network.
This is the single highest-value step you can take. It is also the one that stops malware from spreading sideways, because an intruder that cannot reach the control network cannot disrupt production.
Step 3: Control and monitor the connections you keep
Segmentation does not eliminate the links - it makes them visible and manageable.
- Lock down remote access. Vendor and engineer tunnels should require unique credentials, multi-factor authentication, and be opened only when needed rather than left running permanently.
- Log the boundary. Your firewall between OT and IT should record every allowed and blocked connection. These logs are how you spot a workstation behaving strangely before it becomes an outage.
- Alert on the unusual. A block of denied traffic from a single PC toward the production VLAN at 2 a.m. is worth knowing about the same day, not at month-end review.
- Segment wireless too. Any Wi-Fi on the floor should be its own restricted segment, not an extension of the corporate SSID.
Monitoring is what turns a one-time project into ongoing protection.
Step 4: Harden and patch what you can
Not every device can be updated, but far more can be than most shops assume.
Patch servers, workstations, and network gear on a regular schedule so a known vulnerability on the office side never becomes the door into production. Remove unused software and services. Turn off default accounts and change default passwords on network hardware. Where a device genuinely cannot be updated, compensate with segmentation and monitoring rather than leaving it exposed.
End-user habits matter here too. Most initial infections arrive through email, a downloaded attachment, or a compromised credential. Basic email filtering, endpoint protection, and a staff that knows what a suspicious invoice looks like do more than any single tool.
Step 5: Back up, test, and plan recovery
Segmentation reduces the chance of a production outage. It does not remove it. You still need a way back.
- Back up production data, control configurations, and critical business systems separately, so an event in one zone cannot destroy the other's recovery point.
- Keep an offline or immutable copy so ransomware cannot encrypt the backups along with the source.
- Test restores on a schedule. An untested backup is a hope, not a plan.
- Define recovery time objectives per system. Decide how long each line, server, or application can be down and build the plan around that number.
This is where business continuity work pays off - a documented recovery plan turns a potential week-long shutdown into a controlled, measured response.
Where manufacturing IT solutions fit in
You do not need a full engineering team to do this well, but you do need people who understand both sides of the boundary. That is the difference between generic IT support and manufacturing IT solutions built for production environments - teams who know that a control cabinet is not a server rack and that an unplanned reboot has a cost.
Two services carry most of the load:
- Managed IT support for the office side - monitoring, patching, helpdesk, and vendor management on a predictable monthly fee.
- Business continuity and disaster recovery for the production side - tested backups, defined recovery times, and a documented plan mapped to your critical systems.
Together they cover the boundary in step two and the recovery gap in step five.
Start with the map, not the hardware
Securing OT and IT on the factory floor is not a product you buy. It is a sequence: document the systems, separate the zones, control and log the connections you keep, harden what you can, and prove you can recover.
Most Pennsylvania manufacturers can complete the first two steps inside a normal month without touching production schedules. If you are unsure where your network boundary even sits today, a network assessment is the fastest way to find out - and it costs nothing but a conversation.