Blog

2026.11.10

Factory OT Network VLAN Segmentation: Design and Deployment in Thailand

Factory OT Network VLAN Segmentation: Design and Deployment in Thailand

Deploying VLAN segmentation in a factory OT network starts with understanding the traffic that keeps production running. Configuring VLAN IDs on a switch is only one part of the job. A plant must identify PLCs, HMIs, remote I/O, engineering stations, historians and their actual communication paths, then decide which assets belong in each cell and which flows may cross cell boundaries. This guide explains discovery, VLAN and IP design, routing and access rules, procurement, FAT/SAT and handover for an existing Thailand factory.

The central distinction is simple: a VLAN creates a logical Layer 2 broadcast domain; by itself it is not a complete security boundary. Inter-VLAN routing, ACLs or firewalls, approved traffic rules, logging and change control are needed to govern communication between cells and between IT and OT. Production availability and safety must be tested as part of the same change. ISA/IEC 62443-3-2 describes a process for defining zones and conduits and assessing their risk; it does not establish a universal rule that one VLAN must equal one security zone. See the ISA/IEC 62443-3-2 preview.

What factory OT VLAN segmentation is meant to achieve

An established plant often contains equipment from different years and vendors. PLCs, HMIs, printers, cameras, maintenance laptops and servers may share one flat Layer 2 network because each expansion reused the available switch ports. Eventually no one has a reliable answer to which device is on which port or why a connection exists. Start with one production cell or line whose owner and maintenance window are clear, then use that pilot to validate the migration and rollback process.

VLANs let the designer define the scope of a Layer 2 broadcast domain and some multicast traffic. The Rockwell Automation and Cisco CPwE guide explains that VLANs can group stations by function, line or plant characteristic, while communication between different VLANs requires routing. CPwE Design and Implementation Guide. More VLANs do not automatically mean faster production traffic. Cyclic PLC-to-I/O communication, industrial multicast, time synchronisation and device discovery need protocol-specific and product-specific checks before boundaries move.

A second benefit is a clear place to enforce traffic policy. A production cell may keep its PLC, HMI and I/O together, while historian or MES access crosses an approved Layer 3 boundary. That boundary makes it possible to enumerate and review required flows. If a Layer 3 switch routes every VLAN to every other VLAN without restrictions, logical separation has done little to limit cross-cell access. CISA’s segmentation guidance describes limiting necessary access across OT/IT boundaries using appropriate DMZ and firewall arrangements.

The third benefit is troubleshooting. A documented mapping of cell, port, subnet, gateway and approved destination helps engineers distinguish a cell-local Layer 2 fault from a boundary rule or server outage. VLANs do not supply switch redundancy, power resilience or configuration backup. Those remain separate design tasks. Our related guides on industrial Ethernet switch selection and industrial OT firewall selection cover the procurement choices behind those components.

Build an asset inventory and traffic map before assigning VLAN IDs

The first deliverables are an asset inventory, physical connection diagram, communication flow table and production impact register. Record each switch model, firmware, cabinet and port; connected device and asset owner; MAC and IP address; subnet and gateway; and allowable shutdown window. Check whether installed switches support managed VLAN configuration, 802.1Q tagging, port mirroring, ACLs, IGMP Snooping, logging and configuration backup. When drawings disagree with actual wiring, document the discrepancy and base the migration on the verified connection.

Use passive observation wherever practical on a live control system. A short capture can miss night batches, weekly backup, calibration, month-end transfer or vendor maintenance. Observe representative production modes, changeovers, shutdown and restart, quality inspection and maintenance. If SPAN or TAP is used, check mirror capacity and placement. Active scanning of controllers belongs in an approved test plan agreed with the equipment owner and vendor. NIST SP 800-82 Rev.3 explains why OT controls must account for performance, reliability and safety requirements.

For every required flow, record source, destination, initiating side, direction, protocol, ports or services, frequency or trigger, business purpose, owner and consequence of interruption. “PLC to server over TCP” is too broad to become a firewall rule. Identify the relevant PLC group, server function and application. When a protocol uses dynamic ports, broadcast discovery or multicast, compare vendor documentation with observed traffic. Preserve unknown flows as investigation items instead of automatically approving them.

Discovery itemCheck on siteDesign output
Physical connectionSwitch, port, cable, deviceAccess and trunk port assignment
AddressingIP, subnet, gateway, duplicatesPer-VLAN IP and migration plan
Required flowsPLC–HMI, PLC–I/O, historian, time, maintenanceLocal cell traffic and boundary rules
Operating statesRun, changeover, stop, restart, serviceFAT/SAT test cases
Availability and safetyShutdown consequence, safe behaviour, recoveryMigration window and rollback threshold
Factory OT Network VLAN Segmentation: Design and Deployment in Thailand - figure 1

Distinguish zones, VLANs, subnets and conduits

A zone groups assets with common security requirements; a conduit represents the communication pathway between zones. The published preview of ISA/IEC 62443-3-2:2020 describes partitioning the system under consideration into zones and conduits and documenting their risks and requirements. A VLAN is a Layer 2 logical broadcast domain; an IP subnet is an addressing construct at Layer 3. They may be aligned in an implementation, but they are not interchangeable concepts. Check the ISA standards listing for the version applicable to a procurement.

Imagine cell A containing a PLC, HMI, remote I/O and drives, and cell B containing a second line. Keeping tightly coupled cyclic control traffic within its cell can be practical. Historian and MES traffic can cross an approved boundary. An implementation might assign cell A to VLAN 110, cell B to VLAN 120 and shared OT services to VLAN 300. Those IDs are illustrative, not standard values. If cell A includes safety-related control or an engineering laptop with different requirements, a single-zone assumption may be wrong. Use risk and operational analysis to decide whether another zone, a physical network or device-level control is needed.

An ID such as “VLAN 110” is not a measure of assurance. Each VLAN needs a defined asset scope, switch ports, IP range, gateway, management owner, allowed destinations, prohibited destinations and change procedure. Limit trunk VLANs to those needed. Define handling of unused ports, management interfaces and the native VLAN. Hardware differs, so do not copy one vendor’s configuration into another platform without verification.

Why a VLAN alone cannot enforce the access policy

Inter-VLAN traffic needs routing. An “allow any” policy at that routing point can restore unrestricted communication between nominally separate cells. An immediate “deny everything” policy can instead stop production or maintenance. Build an approved flow table first, then use a Layer 3 ACL or industrial firewall to allow the necessary source, destination and service. Deep packet inspection is an option only if the application and product requirements justify it and tests demonstrate compatibility. The goal is an explicit, maintainable boundary, not the purchase of a particular box.

Traffic between devices in the same VLAN normally bypasses the Layer 3 boundary. Cell composition, port security, device authentication and endpoint configuration therefore remain relevant. Separating IT from OT does not automatically contain all unwanted communication within OT. Layered controls should be selected without introducing untested latency or failure paths into critical control traffic.

Turn the traffic inventory into an approved flow table

An observed flow is evidence of traffic, not proof that it should be permitted. The equipment owner should confirm its purpose. Leave unexplained assets and services pending investigation. Avoid a blanket subnet-to-subnet permission justified only by possible future use. Changes affecting production should be reviewed together by control engineering, operations, maintenance and IT/OT security.

Source zoneDestination zoneExample purposeDesign decision
Cell A PLC/HMICell A I/O and drivesCyclic controlKeep local where required; verify vendor support
Cell A PLCOT historianProduction recordsPermit only approved service and destination
Engineering stationCell A PLCAuthorized maintenanceDefine user, time, path, logs and disconnect
IT business networkCell A PLCRequested direct accessReassess need and design an appropriate path
Cell ACell BInter-line coordinationEvaluate each flow and failure consequence

This table is an illustration, not a list of universally approved flows. Actual ports, initiation direction, multicast, cycle time and failover path depend on the equipment. A historian may poll the PLC, receive a push from it or use a gateway; the permitted direction changes accordingly. Document the initiating party instead of granting two-way access by default.

The ODVA guide on securing EtherNet/IP networks treats VLANs as a way to segment traffic and discusses multicast and IGMP Snooping in the context of a particular network. IGMP Snooping is not a universal cure. The multicast querier, switch implementation, topology and controller requirements must be tested together. The same discipline applies to other industrial protocols whose discovery or real-time behaviour may depend on Layer 2.

Design VLANs, IP addresses and switch configuration together

Create a logical diagram showing cells, shared OT services, maintenance access and the IT/OT boundary. Mark VLAN, subnet and gateway for each segment. Map the logical diagram to a physical drawing of switches, uplinks, ports, cable identifiers, cabinets and power. Link both drawings through the same asset IDs. A logical diagram alone will not reveal a PLC plugged into a switch belonging to another line; a physical drawing alone will not explain the traffic policy.

Decide explicitly whether existing static IP addresses must change. Overlapping IP ranges across VLANs create routing complications. Hard-coded destinations in PLCs, HMIs, applications and maintenance tools must be identified. If IPs change, assess backup files, DNS and NTP settings, historian connections and recovery procedures. If IPs remain, document how routing or mediation will work. Keep a before-and-after address list for acceptance and rollback.

Review access-port VLANs, allowed VLANs on each trunk, management VLAN access, unused ports, error logs and saved configuration. Leaving all VLANs allowed on every trunk may extend a newly created VLAN into an unintended area. Restricting management access is useful, but a badly sequenced rule can lock out the engineering team. Prepare a local console path, tested backup and rollback window before making the change.

The controller vendor’s support conditions matter. Check whether a PLC and remote I/O may be separated, whether an HMI uses Layer 2 discovery, and whether drive or motion traffic relies on a specific multicast pattern. Cisco’s industrial automation design guide describes cell-local cyclic I/O and multicast patterns. Treat reference designs as guidance; validate the exact products, versions and plant topology in service.

Factory OT Network VLAN Segmentation: Design and Deployment in Thailand - figure 2

Migrate a live Thailand plant one cell at a time

A one-night conversion of every production line leaves little room to diagnose hidden flows. Choose a pilot cell with an accountable owner, known maintenance window and manageable traffic. Back up current switch, PLC and HMI configurations. For each change, record purpose, affected ports, operator, outage scope, start and finish times, recovery threshold and approver.

A practical sequence is: survey assets and traffic; test in a lab or stopped equipment; prepare the new boundary; move one cell; verify functions, performance and maintenance access; monitor; then expand. This is not a universal fixed order. Redundant control and safety-related systems require a plan consistent with their product documentation and safety review. Do not run unapproved scans or bulk configuration changes on live control equipment.

Define rollback thresholds before the window begins. “Undo it if communication fails” is too vague. Name the critical PLC–HMI session, cyclic I/O behaviour, quality data path, time source, maintenance access and production limits that determine pass or fail. Agree who can call a rollback. If rollback occurs, preserve the configuration difference and evidence. Do not proceed to the next cell before the cause is understood. A network project should not silently change machine safety functions or fail-safe behaviour.

Pre-cutover checklist

Freeze a consistent version of PLC and HMI backups, switch running and startup configurations, IP list, approved flows, topology and maintenance instructions. Confirm that backup files can be restored to the installed hardware and firmware; simply possessing a file is not a recovery test. Reserve enough of the maintenance window for migration, verification and rollback. Operations and IT/OT teams should use the same acceptance sheet.

Across multiple Thailand sites, standardise asset IDs, naming rules, flow-table format, logging and change approval. Do not force identical VLAN IDs or address blocks where legacy assignments conflict. Align site address plans and remote support paths, and leave documentation that a local engineer can use during an outage.

What to put in an RFP for OT VLAN segmentation

An RFP that says only “separate the OT network by VLAN” may buy switch configuration without discovery, traffic control or testing. Specify the surveyed cells, device and port count, zone proposal, VLAN and IP plan, approved-flow table, switch/L3/firewall configuration, cutover plan, FAT/SAT and operational handover. If the current network is poorly documented, separate a discovery phase from implementation and state how unknown flows affect scope and price.

RFP itemDefine in the requestAcceptance evidence
DiscoveryCells, operating modes, observation period, site accessAsset register and traffic map with exceptions
Logical designVLANs, subnets, gateways, zones and conduitsDrawings, address plan and approved flows
ComponentsInstalled equipment fit, trunk limits, ACL/firewall, logsDatasheets and reviewed configuration
AvailabilityCutover window, failure behaviour, recoveryTest results and rollback procedure
OperationsBackup, change approval, logs, rolesHandover, training and responsibility matrix

Cost depends on more than device count. An unmanaged switch may need replacement; cabinets, power, fibre, cabling, IP changes, PLC/HMI edits, vendor attendance, production downtime, travel, repeated acceptance tests and monitoring can all matter. Compare bids over a common scope, with initial implementation and ongoing licence, support, log retention and change costs separate. There is no defensible universal price for a plant whose installed network and shutdown constraints have not been surveyed.

State that configuration is not accepted merely because VLAN IDs exist. Required flows must work in relevant operating modes; unapproved cross-cell flows must be denied; operators must be able to interpret logs; management access must be limited; and configuration recovery must be demonstrated. Obtain process-specific performance and safety criteria from the equipment owner rather than importing generic internet thresholds. Record each exception with owner, reason, expiry and review date.

FAT and SAT: test both working and blocked traffic

At FAT, check port VLANs, allowed trunks, routing, ACL or firewall rules, logs and configuration restore in a lab or pre-delivery environment. Where possible use actual PLC/HMI hardware or equivalent test equipment to prove application traffic, not only ICMP ping. Mark tests that cannot be completed without the plant as open SAT items.

At SAT, use actual site wiring, addresses and production settings. Verify PLC–HMI display and commands, I/O response, historian data, recipes, quality records, NTP, engineering access, alarms and higher-level systems in the relevant modes. Also test recovery events that are part of the adopted design: controller restart, switch replacement, upstream failure and maintenance access opening and closure. The equipment owner must approve acceptance values for cycle timing and allowable outage; these differ by product and process.

Negative testing matters too. Confirm that an unauthorized host in cell B cannot access cell A’s PLC, that the IT business network has no unintended direct route into a control cell, and that unused ports are handled as specified. Do not subject live controllers to unapproved load or attack-like tests. Use an isolated lab or a carefully approved site method. Verify that logs identify source, destination, rule and time so an operator can investigate a denial.

Factory OT Network VLAN Segmentation: Design and Deployment in Thailand - figure 3

Five frequent implementation mistakes

1. Choosing VLAN IDs before mapping traffic. Discovery, cyclic I/O or historian communication stops after cutover. Build the asset and flow map first and check the controller vendor’s specifications.

2. Allowing every VLAN to route to every other VLAN. Broadcast domains are separated but cross-cell access remains open. Approve the flow table, enforce it with ACLs or a firewall and test both positive and negative cases.

3. Treating broadcast, multicast and discovery as one thing. IGMP Snooping alone cannot prove correct behaviour. Check the querier, topology, switch implementation and application protocol with real equipment.

4. Trusting old drawings and IP lists. Hidden connections or duplicate addresses appear during the change. Verify physical ports, static destinations, gateways, servers and the before-and-after address list.

5. Leaving recovery with the installer. A night-shift maintenance team cannot restore service. Handover a tested backup, restore procedure, contact list, drawing, rule register and change history with clear ownership.

Keep the segmentation design current after go-live

A VLAN project is not complete on the day the switch configuration is applied. Each new machine, replacement PLC, changed engineering laptop or migrated server should trigger an update to port mapping, addressing, approved flows, access owner, logs and backup. Exceptions without expiry erode the original design. Change requests should say why a flow is needed, which rule is added or removed, what equipment is affected, how it was tested and how it can be rolled back.

Useful operational measures include unexpected denials, production interruption, false alarms, unidentified ports, expired exceptions and backup success. Targets should be set for the actual plant and monitoring capability. Do not widen a rule merely because it produces many deny events; determine whether they are missed requirements, unnecessary traffic or device faults. Preserve the performance, reliability and safety considerations emphasised in NIST’s final OT security guide.

Version control matters for sources as well. As of November 2026, this article refers to the final NIST SP 800-82 Rev.3, the published preview of ISA/IEC 62443-3-2:2020, the ISA standards list and vendor design documentation. NIST has issued an initial public draft of Rev.4, which is not treated here as a final requirement. Check the applicable standard edition and product specification again when issuing an RFP.

Summary

Factory OT VLAN segmentation works best when it begins with actual assets and traffic, then combines VLANs, IP design, routing and explicit access control. Test both necessary control traffic and unnecessary cross-cell traffic, plus recovery, at FAT/SAT. A properly accepted pilot cell and maintained change register provide a sound basis for expansion across a plant.

If you are considering VLAN segmentation around existing PLCs and industrial switches in a Thailand factory, you can contact TOMAS TECH while you are still mapping the equipment. We can help define the discovery scope, approved traffic table, phased migration and acceptance criteria.

FAQ

Does separating factory OT VLANs automatically make the network secure?

No. VLANs provide Layer 2 logical separation. Inter-VLAN routing requires explicit ACL or firewall policy, while devices within one VLAN need their own access review. Logging, maintenance access and change control are also part of the design.

Must every equipment cell have exactly one VLAN?

No. Decide based on the control protocol, traffic, ownership, risk, physical topology and vendor support. A zone may use more than one VLAN, and assets with different risks may need distinct zones.

Must existing PLC IP addresses change?

Not necessarily. Check the relationship between VLANs and subnets, routing, fixed destinations and duplicate addresses. If IPs do change, update HMIs, historians, maintenance tools, backups and the rollback plan.

What should a VLAN segmentation FAT/SAT accept?

Required application flows, blocked unapproved flows, effect on control timing and production, recovery, logs, backup and local maintenance procedures. Ping alone is insufficient; numerical pass criteria come from the equipment and process.

Should we select a firewall before defining VLANs?

First identify assets, zones, necessary communication and outage consequences. Then design VLANs and the Layer 3 boundary and choose the control functions needed there. Product selection before flow discovery often leads to unsuitable scope.

Primary references