The short answer
Industrial Firewalls
An industrial firewall is a packet-filtering device hardened for a control environment: DIN-rail or rack mounting, 24 V DC power, an extended temperature range, no fans, a long product life, and rule sets that understand industrial protocols so that a rule can permit Modbus reads and deny Modbus writes. They sit at the boundaries between zones, at the DMZ edges, in front of individual controllers that cannot protect themselves, and at remote sites on the telemetry link, in transparent mode where the network cannot be re-addressed and routed mode where it can. They are managed like any network device, with backed-up configurations, named accounts, and logs that someone reads.
Key points
- Industrial firewalls are ruggedized, protocol-aware, and built to be installed in a panel and left for a decade.
- They enforce zone boundaries: DMZ edges, between plant zones, in front of legacy controllers, and at remote sites.
- Transparent mode filters without changing addresses; routed mode separates subnets. Both have their place.
- Protocol awareness lets a rule allow reads and block writes or program downloads.
- A firewall nobody manages is a switch. Configuration backup, accounts, updates, and log review are the job.
What is different about an industrial firewall
| Property | Industrial firewall | Office or data center firewall |
|---|---|---|
| Form | DIN-rail or small rack unit, 24 V DC, fanless, wide temperature range, vibration rated | Rack unit, AC power, fans, controlled environment |
| Lifecycle | Ten years or more of support; firmware maintained for the installed base | Three to five years; frequent model changes |
| Protocols | Deep packet inspection for Modbus, DNP3, EtherNet/IP, OPC UA, and others; rules by function code and register | Web, email, and enterprise application awareness |
| Modes | Transparent bridge or routed, often both; some act as a Layer 2 filter with no IP address of their own | Routed, with NAT and VPN as the norm |
| Throughput | Modest; control traffic is small | High; designed for internet-scale traffic |
| Management | Local web page, a central management tool, or a controller-vendor ecosystem | Enterprise management platforms |
| Failure mode | Configurable: fail closed, or fail open where the process cannot tolerate a stop | Fail closed |
The office firewall at the enterprise edge stays where it is. Industrial firewalls are for the boundaries inside and around the control system, where a device in a panel at a lift station has to survive the panel and the traffic is a few Modbus polls a second.
Where they go
- The DMZ boundaries
- The two boundaries of the industrial DMZ, enterprise to DMZ and DMZ to control, are firewalls. The enterprise side may be the enterprise firewall; the control side is usually an industrial unit or a control-network-owned enterprise-class unit. The DMZ design page covers the rules.
- Between plant zones
- Where the zone drawing puts a boundary that must be enforced: between the SCADA server zone and the plant control zone, between the engineering zone and the controllers, around a vendor package.
- In front of a legacy controller
- A single small firewall on the controller network port, allowing only the SCADA master and the engineering workstation on the protocol ports, with writes and downloads restricted. The compensating control for a controller with no security of its own.
- At a remote site
- Between the telemetry radio or cellular modem and the site controller, allowing only the SCADA protocol from the master and blocking everything else, including site-to-site traffic. Often built into the cellular router.
- Around a cellular or vendor connection
- Wherever a link the utility does not fully control enters the network.
Transparent or routed
A transparent firewall bridges two network segments at Layer 2 and filters the frames passing through it; the devices on both sides keep their addresses and do not know the firewall exists. It is installed by cutting a cable, which makes it the practical choice for protecting an existing controller or a segment that cannot be re-addressed. A routed firewall separates two subnets and forwards between them according to its rules; it is the choice at a zone boundary that the network design already treats as a separate subnet, and it can also do NAT and VPN. Many industrial units do either, and a utility network commonly has both: routed at the zone boundaries, transparent in front of individual devices.
Protocol awareness
A conventional firewall rule permits or denies by address and port. Modbus TCP is port 502; a rule that permits port 502 permits every Modbus function, including writes to holding registers and coils. An industrial firewall with deep packet inspection reads the Modbus function code and register range inside the packet, so the rule can say: permit function codes 3 and 4, reads, from the SCADA server to this controller for registers 0 to 999, and deny everything else. The same applies to DNP3 function codes, EtherNet/IP services, and controller programming protocols, so that a program download is permitted only from the engineering workstation and only when a session is enabled. This is the capability that turns a firewall from a fence into a policy.
Managing them
- Configuration backed up after every change and stored with the other network configurations.
- Named administrator accounts, MFA where the device supports it, default credentials removed, and the management interface reachable only from the engineering zone.
- Firmware updated on a schedule, tested on a spare, because a firewall vulnerability is a hole in the boundary.
- Rules documented against the conduit table, each with its conduit and its reason, reviewed on a schedule.
- Logs sent to a log server, with denied traffic reviewed and permitted traffic sampled.
- A spare on the shelf with the configuration loaded, because a failed firewall at a remote site is a site without telemetry.
- On the network drawing, with its addresses, its mode, and its rule set reference.
Selecting one
The choice is driven by where it goes: a DIN-rail unit with two or four ports for a panel or a remote site, a rack unit with more ports and higher throughput for a zone boundary or the DMZ. The protocol support must cover what the conduit carries, the management must fit how the utility manages its other network devices, and the vendor must have a record of supporting the product for the life of the panel it goes in. Controller vendors offer firewalls integrated with their programming ecosystems, network vendors offer general industrial units, and both work; a utility standardizes on one or two so that its staff and its spares cover them.
Frequently asked questions
- Do we need a firewall in front of every PLC?
- In front of the ones that cannot protect themselves and that matter: controllers with no authentication on a network that other things can reach. A plant with all its controllers in one zone behind a zone firewall, with no other devices in that zone, may not need per-device firewalls. A controller sharing a network with workstations or vendor equipment does.
- Can the cellular router at a remote site be the firewall?
- Most industrial cellular routers include a stateful firewall and VPN, and at a small site that is the firewall. Configure it: allow the SCADA protocol from the master only, deny everything else inbound, VPN to the utility concentrator, no management from the cellular side, default credentials changed, and the configuration backed up.
- What throughput do we need?
- Very little. A lift station generates kilobits per second; a plant control zone generates a few megabits. Industrial firewalls are rated in the hundreds of megabits and the constraint is never throughput. Port count, mode, protocol support, and environment are the constraints.
- Should the firewall be the same brand as the controllers?
- Not necessarily. A controller vendor firewall integrates with that vendor tools and understands that vendor protocols well; a network vendor unit understands more protocols and fits a mixed plant. The zone firewalls and the DMZ firewalls are often better from a network vendor, and the per-controller units from either. Mixing vendors at the two DMZ boundaries is a deliberate practice.
Related topics
- Zones and ConduitsThe IEC 62443 way to segment a control system: grouping assets into zones with a shared security level, inventorying every conduit between them, and turning the drawing into firewall rules. With a worked water utility example.
- Industrial DMZ DesignThe buffer zone between the business network and the control system: what goes in it, the no-direct-path rule, push-not-pull data flows, the firewall pair, and the services a utility actually needs to place there.
- Firewall Rule DesignWriting the rule set for a control network boundary: default deny, one rule per conduit with its reason, hosts not subnets, protocol and port and function, direction and initiator, logging on every rule, and the review that removes forgotten rules.
- Controller HardeningWhat can be locked down on a PLC or RTU: keyswitch and mode, credentials, unused services, access lists, firmware, and the engineering workstation that is the real target.
- Jump HostsThe one machine every remote session must pass through: what a jump host is, how it is hardened, what it may reach, session recording, the tools it carries, and why a vendor laptop never gets past it.
- What to Back Up in a Control SystemWhat a utility must be able to restore after a failure or attack: controller programs and firmware, SCADA and historian, HMI panels, drive and instrument parameters, network configurations, licenses, installers, and documentation, and how often each changes.
Direct contact
Have a controls question?
Reach Eric Sullivan directly about anything on this site, a controls or automation topic, or one of his personal projects.