What is Topology?
Topology, or network topology, is the way devices and connections in a network are physically and logically arranged, and therefore determines the path data takes between those devices. In industrial and IT networks this concerns the links between PLCs, sensors, HMIs, servers and field components. The chosen shape, such as star, line, ring or tree, determines how much cable you need, what happens when a cable breaks, how quickly data arrives and how easily you can extend the network later.
🧠 What is the difference between physical and logical topology?
Every network has two topologies, which do not have to be the same:
- Physical topology - how cables, switches and devices are actually connected; this is what you see in the cabinet and on the cabling drawing
- Logical topology - how data behaves as it moves through the network, regardless of the cabling
| Aspect | Physical topology | Logical topology |
|---|---|---|
| Definition | How devices are physically connected | How data flows behave |
| Example | Star around a switch | Logical ring in EtherCAT: the frame passes every device and the last device sends it back |
| Important for | Cabling, maintenance, troubleshooting | Network behaviour, protocols, fault detection, latency |
| Recorded in | Cabling plan, cabinet drawings | Protocol configuration, VLANs, redundancy protocols |
Two examples show the difference. EtherCAT is often physically a line but logically a ring, because the frame returns to the master over the same cables. Conversely, a physical ring with MRP is logically a line, because the redundancy protocol blocks one port until a cable breaks.
🔧 Which network topologies are most common?
| Topology | Description | Advantages | Disadvantages | Typical in OT |
|---|---|---|---|---|
| Bus | All devices are attached to one shared medium | Little cable, inexpensive | A cable fault or poor termination affects the whole segment | RS-485, PROFIBUS DP, Modbus RTU |
| Line (daisy chain) | Devices with a built-in switch are chained one after another | Simple, short cables along a machine | A break cuts off every device behind it; each device adds delay | PROFINET, EtherCAT, IP |
| Star | Each device has its own cable to a central point (switch) | Clear layout, one cable fault affects one device | The central point is a single point of failure, more cable | Office IT, control networks |
| Ring | A line whose ends are closed | Survives a single break | Requires a ring protocol, otherwise it forms a loop | MRP, DLR, HSR |
| Tree | Hierarchy of stars or lines on a backbone | Scalable, follows the plant layout | A failure high in the tree affects many devices | Plant backbone with cells |
| Mesh | Devices have several connections to each other (partial or full) | Very high availability, multiple paths | Expensive and complex, needs routing or RSTP | WirelessHART, core networks |
The classic bus also existed in Ethernet (coaxial cable, 10BASE5 and 10BASE2), but switched Ethernet is always physically a star, line or ring. In OT you mainly find a true bus in fieldbuses.
🏭 Which topology do industrial protocols use?
In industrial automation, the protocol often already determines which topologies are possible:
- AS-Interface - free topology (line, star, tree and also ring) without a terminating resistor, using an unshielded flat cable; a maximum of 100 metres per segment, extendable with extension plugs and repeaters, and up to 62 slaves per network
- PROFIBUS DP - a line (bus) on RS-485 with a maximum of 32 stations per segment, actively terminated at both ends; with repeaters up to 126 addresses can be reached. Segment length drops from 1,200 metres at low speed to 200 metres at 1.5 Mbit/s
- EtherCAT - line, tree, star or daisy chain; up to 65,535 devices per segment and a maximum of 100 metres of copper between two devices. A ring is used only for cable redundancy
- PROFINET and IP - star and tree via industrial switches, or a line via devices with two ports; a ring with MRP (PROFINET) or DLR (EtherNet/IP)
- Ethernet-APL - a trunk of up to 1,000 metres with spurs of up to 200 metres to the field instruments; the trunk itself does not support a ring, redundancy sits in the backbone
- WirelessHART - a wireless mesh network in which every instrument forwards messages from its neighbours
In a long line, every chained device adds to the delay. The greater the line depth between controller and device, the longer the update time the engineering tool has to choose. This matters for deterministic and real-time behaviour.
🔄 How does topology provide redundancy?
A star or a line has single points of failure by definition. For installations that must not stop, you therefore choose a topology with a second path, combined with a protocol that prevents loops:
| Protocol | Standard / origin | Topology | Recovery time |
|---|---|---|---|
| MRP | IEC 62439-2 | Ring | 10, 30, 200 or 500 ms (class); 200 ms for up to 50 devices |
| DLR | ODVA | Ring of devices | Less than 3 ms with 50 nodes |
| RSTP | IEEE 802.1w | Any topology, including mesh | Usually sub-second, not guaranteed |
| PRP | IEC 62439-3 | Two separate parallel networks | 0 ms (seamless) |
| HSR | IEC 62439-3 | Ring, every frame sent in both directions | 0 ms (seamless) |
You can read more about the trade-offs under ring redundancy and redundancy. Important: a ring without a working redundancy protocol is a loop, and a loop causes a broadcast storm that brings down the whole segment.
🎯 How do you choose the right topology?
A practical approach in five steps:
- Define the availability requirement - can a cell drop out for a few seconds, or must recovery stay within the I/O watchdog time? That determines whether a star is enough or whether a ring or PRP is needed.
- Look at the protocol and the devices - which topologies does the protocol support, and do the devices have two ports with a built-in switch?
- Record the physical limits - 100 metres per copper link with standard Ethernet, fibre for longer distances or between buildings, and the segment limits of fieldbuses.
- Follow the plant layout - a tree with its own ring or star per production cell fits the Purdue Model and makes segmentation easier.
- Document and test - record the topology in drawings and in the asset inventory, and during commissioning pull a cable at several points to test the failover.
An example: a packaging line with ten PROFINET devices, each with two ports, can be cabled as a single line. If you close that line into an MRP ring with one extra cable, the machine survives a cable break, provided every device supports MRP and one switch or controller takes the ring manager role. The extra cost is then often limited to that one additional cable.
The installation of industrial networks is covered by IEC 61918, which includes an informative annex on network topologies, supplemented by protocol-specific installation profiles in the IEC 61784-5 series. Work with these typical values:
| Parameter | Typical value |
|---|---|
| Copper (100BASE-TX) | Maximum 100 m per link |
| RS-485 segment | Maximum 32 stations, up to 1,200 m at low speed |
| AS-Interface segment | 100 m without extension |
| Ethernet-APL | Trunk up to 1,000 m, spur up to 200 m |
| MRP ring (200 ms class) | Maximum 50 devices |
🔐 What does topology mean for OT security?
Topology determines where traffic passes, and therefore where you can filter and monitor it. A flat star in which everything talks to everything makes it easy for an attacker to move laterally. Under the zones and conduits model of IEC 62443, you group devices into zones and let traffic pass only through controlled conduits, such as an industrial firewall between a cell ring and the backbone.
- Know your topology - an up-to-date network drawing is the basis for asset discovery and network monitoring
- Monitor rings - redundancy protocols such as MRP and DLR have no authentication; an unexpected failover can indicate a fault or sabotage
- Limit broadcast domains - split large layer 2 networks with VLANs and routers, so that a fault or an attack does not hit the whole plant
❓ Frequently asked questions
What is the best topology for an industrial network?
There is no single best network topology for every situation. In practice, plants combine a redundant backbone (ring or mesh) with a ring, line or star per production cell, depending on the protocol and the required availability. The choice follows from the availability requirement, the distances and the devices you use.
What is the difference between a bus and a line topology?
In a bus topology all devices share one electrical medium, such as an RS-485 cable in PROFIBUS DP. In a line topology, devices with a built-in switch are chained one after another, as in PROFINET or EtherCAT, and every connection is a separate point-to-point link. A cable break in a line affects only the devices behind the break.
Why does a ring topology need a redundancy protocol?
In Ethernet, a closed ring forms a loop in which broadcast frames circulate endlessly, causing a broadcast storm. A redundancy protocol such as MRP or DLR therefore blocks one port and opens it only when a cable breaks. The ring then works logically as a line and physically as a backup path.
How long can an Ethernet cable be in an industrial network?
A twisted-pair copper Ethernet link such as 100BASE-TX can be at most 100 metres long between two active devices. For longer distances or links between buildings you use fibre. Ethernet-APL is an exception and reaches up to 1,000 metres on the trunk.
What is a logical ring in EtherCAT?
EtherCAT is usually physically a line but logically a ring: the master sends one frame that every device reads and writes on the fly as it passes. The last device detects an open port and sends the frame back to the master over the return channel of the full-duplex link. As a result, EtherCAT does not need a separate return cable.
What role does topology play in network segmentation?
Topology determines where you can place zones and conduits. A tree with its own switch or ring per production cell makes it easy to put a firewall between the cell and the backbone, while a flat network without hierarchy makes segmentation difficult. You should therefore design network topology and segmentation together.
📌 In summary
Topology describes how devices in a network are physically and logically connected, and therefore determines cost, latency, expandability and what happens when something fails.
In OT the protocol often determines the possible shapes: bus for fieldbuses, line and ring for industrial Ethernet, mesh for wireless instrumentation. Choose the topology based on the availability requirement, distances and scalability, always combine rings with a redundancy protocol, and design topology and segmentation together.