Fiber Optic Network Design for Industrial Plants: Ring vs Star Topology
Key Takeaway
Choosing the right fiber optic network topology is a critical decision that directly affects plant uptime, recovery time, and long-term scalability. This guide compares ring and star topologies, covers ring redundancy protocols including RSTP, MRP, and DLR, and provides fiber count planning and architecture guidelines following Cisco CPwE and Rockwell IACS-E028 standards.
Why Topology Matters for Plant Uptime
In an industrial plant, a communication network outage is not merely an inconvenience — it can mean a production line shutdown, a safety system relying on stale data, or an operator unable to respond to an alarm. The physical topology of the fiber optic network — how cables are routed and how switches are interconnected — determines how the network responds to a fiber cut, a switch failure, or an accidental cable pull. Getting topology right during the design phase costs almost nothing compared to retrofitting redundancy after an outage event.
Two fundamental topologies dominate industrial fiber networks: star (hub-and-spoke) and ring. Each has legitimate applications, and many plants use a combination. The decision involves uptime requirements, the criticality of each node, fiber cost, and the redundancy protocol capabilities of the installed switches.
Star Topology: Advantages and Limitations
In a star topology, all edge switches connect back to a central or distribution switch by independent point-to-point fiber links. Each link is independent, so a single fiber cut affects only the one device connected to that link. Star topology is simple, easy to troubleshoot, and uses standard Ethernet switching without any special redundancy protocol configuration.
Star topology has two significant limitations in industrial plants. First, it creates a single point of failure at the central or distribution switch: if that switch fails or its fiber uplink fails, all devices it serves lose communication simultaneously. Second, running individual fiber home-runs from each field device to the central switch consumes more fiber and conduit than ring topology for sites where field devices are geographically close to each other.
Star topology is appropriate for:
- Non-critical monitoring applications where intermittent data loss is acceptable
- Building automation systems in ancillary plant buildings
- Small plants with all equipment within 100 meters of the central switch
- Systems with hardware redundancy at the device level that do not depend on network continuity
Ring Topology: Protocols and Recovery Times
In a ring topology, switches are connected in a closed loop. Each switch has two fiber uplinks — one to the next switch clockwise and one to the next switch counter-clockwise. Under normal operation, a protocol blocks one of the links to prevent a broadcast loop. When a fiber cut or switch failure occurs, the blocking is removed and traffic reroutes around the break within the protocol's recovery time.
Three ring redundancy protocols are commonly deployed in industrial plants, each with different recovery time characteristics:
RSTP/MSTP — Spanning Tree
Rapid Spanning Tree Protocol (RSTP, IEEE 802.1w) and Multiple Spanning Tree Protocol (MSTP, IEEE 802.1s) are standard Ethernet protocols available on virtually all managed industrial switches. RSTP recovery time after a link failure is typically 1 to 30 seconds depending on network size and switch configuration. For process control applications with scan rates of 100ms to 500ms, RSTP recovery may cause a brief but detectable control interruption. RSTP is acceptable for SCADA monitoring applications where data continuity during the recovery window is not critical to safety or process stability.
MRP — Media Redundancy Protocol
Media Redundancy Protocol (IEC 62439-2) provides sub-50ms ring recovery, making it suitable for real-time process control. MRP operates in a ring of up to 50 switches (IEC limit; practical rings are typically 10 to 20 switches for performance reasons). One switch in the ring is designated the Media Redundancy Manager (MRM) and monitors ring health. When a link fails, the MRM opens the blocked port and sends frames to flush forwarding tables in adjacent switches, allowing traffic to resume within 10 to 50ms. MRP is supported by Siemens, Phoenix Contact, Hirschmann, Moxa, and many other industrial switch manufacturers. Vendor-specific MRP variants (Hirschmann HIPER-Ring, Siemens HRP) offer sub-30ms recovery.
DLR — Device Level Ring
Device Level Ring (DLR) is a Rockwell Automation protocol for EtherNet/IP networks using ControlLogix and CompactLogix PLCs. Unlike MRP, which operates at the switch layer, DLR is implemented at the end-device level in Rockwell I/O modules, drives, and other CIP devices. A DLR ring supervisor (typically a ControlLogix with a 1756-EN2TR dual-port Ethernet module) monitors ring health and initiates recovery in under 3ms — the fastest recovery time of any common industrial ring protocol. DLR is ideal for high-speed motion control and safety PLC networks where even 50ms of communication loss is unacceptable. DLR requires all devices on the ring to support DLR; mixing non-DLR devices on a DLR ring is not permitted.
Protocol Comparison
| Protocol | Recovery Time | Standard | Best Use Case |
|---|---|---|---|
| RSTP | 1–30 seconds | IEEE 802.1w | Monitoring, non-critical SCADA |
| MSTP | 1–30 seconds | IEEE 802.1s | Multi-VLAN networks, large plants |
| MRP | <50ms | IEC 62439-2 | Process control, DCS, PLC networks |
| DLR | <3ms | Rockwell/ODVA | EtherNet/IP, motion, safety PLCs |
Fiber Counts Per Link
Every point-to-point fiber link should carry a minimum of two fibers: one for transmit and one for receive. Single-fiber operation using bidirectional (BiDi) SFPs is not recommended for industrial applications because a single-fiber failure causes complete link loss rather than a graceful switch to a backup path.
For ring topology links, install a minimum of four fibers per link segment: two fibers for the active link and two dark fibers for future redundancy or to support additional network segments without new cable installation. On backbone runs between buildings or major distribution points, six or more fibers per link is standard practice to accommodate future growth, OTDR monitoring, and spare capacity.
Distribution Architecture: OSP Backbone Ring to Building Stars
Large industrial plants typically deploy a two-tier fiber architecture following Cisco's Converged Plantwide Ethernet (CPwE) and Rockwell's IACS-E028 guidance:
- Plant backbone ring (OSP layer): A fiber ring connecting the main control room, substations, MCCs, and major process buildings using single-mode OS2 fiber. Distribution switches at each building serve as the ring nodes. This ring uses MRP for sub-50ms recovery.
- Building or area star networks (ISP layer): From each distribution switch, star connections run to field panel switches, operator workstations, and I/O panels within the building or process area. These connections use multimode OM4 fiber for runs under 400 meters or OS2 for longer runs.
This architecture limits the blast radius of any single failure. A fiber cut on the OSP backbone is recovered by MRP in under 50ms. A failure at a building distribution switch affects only that building, not the entire plant network.
Patch Panel Placement Strategy
Patch panels provide a demarcation point between the permanent fiber plant (spliced backbone cables) and the active equipment connections (patch cords). Proper patch panel placement simplifies troubleshooting and reduces the risk of accidental cable damage during equipment maintenance.
- Control room: A main distribution frame (MDF) with patch panels for all backbone fiber terminations. Patch cords connect backbone panels to core switches. 24-port or 48-port panels organized by plant area.
- Field junction boxes: Weather-rated enclosures at process area boundaries with 6- to 12-port panels for local fiber terminations. Reduce the number of individual fiber runs back to the control room.
- MCC rooms: Dedicated fiber panels for drives, soft starters, and motor protection relays connected via EtherNet/IP or Modbus TCP. Typically 6- to 12-port panels mounted in a dedicated fiber management section of the MCC.
Fiber Count Planning
A fundamental rule in industrial fiber design: install a minimum of four times the current fiber requirement at each location. The marginal cost of additional fiber during installation is small — fiber cable cost scales modestly with fiber count — while the cost of installing additional cable after the fact includes trench work, conduit occupancy analysis, pulling operations, and production impact from potential outages during work.
Fiber count planning should account for:
- Current active connections (2 fibers per active link)
- Ring redundancy (2 additional fibers per ring link)
- Future network expansion (one-third of total count as dark spares)
- OTDR monitoring fibers (1 fiber per segment for continuous monitoring)
- Potential OT/IT convergence requirements (separate fiber pairs for IT traffic)
Network Zoning: OT vs IT Fiber Separation
Industrial cybersecurity frameworks including IEC 62443 and NIST SP 800-82 require logical and, where practical, physical separation between OT (operational technology) networks and IT (corporate/business) networks. In fiber optic infrastructure, this means maintaining separate fiber pairs for OT traffic and IT traffic on shared backbone cables, or installing physically separate cables for each network zone.
Physical separation of OT and IT fiber is the most secure approach and eliminates the risk of VLAN misconfiguration bridging the two networks. Where cable cost requires a shared physical cable, documented separate fiber pair assignments and independent patch panels for OT and IT should be maintained.
Switch Placement and Fiber Port Types
Industrial switches use SFP (Small Form-factor Pluggable) modules to terminate fiber connections. SFP module selection must match the fiber type and required reach:
| SFP Type | Fiber | Wavelength | Maximum Reach |
|---|---|---|---|
| SX (Short Reach) | OM2/OM3/OM4 | 850nm | 550m (OM4) |
| LX (Long Reach) | OS1/OS2 | 1310nm | 10km (OS2) |
| ZX (Extended Reach) | OS2 | 1550nm | 80km |
| BiDi SX | OM4 | 850/910nm | 300m — not recommended for industrial rings |
All SFPs on a given link must match in type, wavelength, and connector style (LC is standard for SFP; SC adapters are available for legacy panels). Mixing SFP types on a link — for example, connecting an SX SFP to an LX SFP — will prevent the link from establishing.
NFM Consulting Fiber Optic Services
NFM Consulting designs and installs industrial fiber optic networks following Cisco CPwE and Rockwell IACS-E028 architecture guidelines. Our network engineers produce complete fiber design packages including topology diagrams, fiber count schedules, conduit routing drawings, and switch configuration specifications. We perform OTDR certification testing on every link and provide as-built documentation. Contact NFM Consulting to discuss network topology design for your industrial plant.
Frequently Asked Questions
RSTP (IEEE 802.1w) recovery time in a ring is typically 1 to 30 seconds depending on the number of switches and network configuration. MRP (IEC 62439-2) recovers in under 50ms — roughly 20 to 600 times faster than RSTP. For process control applications with 100ms PLC scan rates, RSTP recovery can cause a detectable control gap while MRP recovery is transparent to control processes. Vendor-specific MRP variants like Hirschmann HIPER-Ring and Siemens HRP can achieve sub-30ms recovery.
Install a minimum of four fibers per ring segment: two for the active link (transmit and receive) and two dark spares. On backbone runs between buildings, six or more fibers per segment is recommended to accommodate future network expansion, OTDR monitoring, and additional network segments without requiring new cable installation. The marginal cable cost for additional fibers is small compared to the cost of future cable replacement.
DLR (Device Level Ring) is preferred when the EtherNet/IP network includes high-speed motion control, safety PLCs, or I/O with scan rates under 10ms where even 50ms of communication loss would cause a fault or process upset. DLR achieves recovery in under 3ms by implementing redundancy at the device level rather than the switch level. DLR requires all devices on the ring to support the DLR protocol — verify support in the Rockwell Product Compatibility and Download Center (PCDC) for all I/O and drive models before specifying DLR.
Physical separation is preferred from a cybersecurity standpoint and satisfies IEC 62443 defense-in-depth requirements. When cable cost requires sharing, the OT and IT networks should use separate fiber pairs within the same cable, with separate patch panels, separate switch ports, and no shared patch cords. Never bridge OT and IT VLANs at the switch level, as this creates a lateral movement path between networks in the event of a security incident.
The two most common SFP types for industrial plants are SX (850nm, OM3/OM4 multimode, up to 550m) for intra-building connections and LX (1310nm, OS2 single-mode, up to 10km) for inter-building and cross-plant runs. Both use LC connectors on standard SFP modules. Always verify that both ends of each link use identical SFP type, wavelength, and connector — mismatched SFP types on a link will prevent the link from establishing, even if the physical connectors mate.