In a plain daisy-chained or star network, one cut cable or one dead switch can silence a whole string of devices. On a facility where those devices control real equipment, that is unacceptable. Network redundancy is how engineers build a network that survives a single failure - most commonly with a ring topology that heals itself. This guide explains how it works and why field networks use it.
Network Redundancy (Ring) in one line: Network redundancy is designing a network so that no single link or device failure disconnects any part of it, by providing more than one path between points. The most common industrial form is a ring topology, where switches are cabled in a loop and a recovery protocol keeps one link idle until a failure occurs, then activates it to heal the network automatically.
A ring cables the switches in a closed loop, so between any two points there are two possible paths around the ring. If one cable is cut or one switch fails, traffic simply flows the other way around, and no device is isolated. Compared with a star (every device homed to one central switch, a single point of failure) or a daisy chain (one break splits the string), the ring's second path is what provides the resilience.
But a physical loop creates a serious problem on its own: broadcast frames circulate endlessly, multiplying until they saturate the network - a broadcast storm that takes everything down. A ring therefore cannot simply be closed and left. A protocol must logically block one link so the topology is loop-free in normal operation, and then unblock it the instant a real failure breaks the loop elsewhere. Managing that block-and-heal behaviour is the whole job of a ring-recovery protocol.
The standard mechanism is the Spanning Tree Protocol and its faster successor, Rapid Spanning Tree Protocol (RSTP). RSTP computes a loop-free tree across the switches, deliberately blocking redundant links, and when a link fails it recalculates and brings a blocked link into service. RSTP recovery is typically a few seconds - fine for many networks but too slow for time-critical control, where a multi-second gap can trip a process.
For that reason, industrial switch vendors offer proprietary ring protocols that recover far faster, often in tens of milliseconds, fast enough that control traffic barely notices the break. These trade the openness of RSTP for speed and are common on control-critical rings. The choice comes down to recovery time: RSTP where a few seconds is acceptable, a fast vendor ring where the process cannot tolerate that gap. Either way, the principle is the same - keep a spare path blocked, and switch to it the moment the primary path fails.
On a larger facility - a gas plant, compressor station, or big pad - the local OT network often carries control traffic among many PLCs, RTUs, and flow computers, and a single cable fault should not blind an operator to a whole section. Wiring the managed industrial switches in a self-healing ring gives that network a second path, so a backhoe, a rodent, or a failed port heals automatically instead of taking the site down until someone finds the fault. Redundant power to the switches complements the redundant paths.
Redundancy lives in the network layer, below the data. It does not create tags; it keeps the path to them alive. A cloud SCADA such as Merobix polls the controllers reachable across that ring over Modbus TCP, EtherNet/IP, DNP3, or OPC UA, and a self-healing ring is what keeps that polling uninterrupted when a single link fails. On a site where visibility into the process is safety-relevant, network redundancy is what turns a potential outage into a non-event.
A ring cables the switches in a closed loop so there are two paths between any two points. If one link or switch fails, traffic flows the other way around the ring and no device is isolated. A recovery protocol keeps one link blocked in normal operation to prevent a loop.
With Rapid Spanning Tree Protocol (RSTP), recovery is typically a few seconds. Proprietary industrial ring protocols recover much faster, often in tens of milliseconds, which is needed for time-critical control traffic that cannot tolerate a multi-second gap.
An unmanaged physical loop causes a broadcast storm - frames circulate endlessly and saturate the network. You need a managed switch running a ring or spanning-tree protocol that logically blocks one link until a failure occurs, then activates it. The protocol is what makes a loop safe and self-healing.
This page references the protocol specifications published by the organizations below. Editions, product capabilities, and documentation change over time - confirm current requirements and specifications directly with the source.
Last reviewed: July 27, 2026. Merobix is not affiliated with, endorsed by, or sponsored by these organizations; their names are used only to identify the standards and products discussed.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.