What Is Ring Topology?

Ring topology is a network layout in which every device connects to exactly two others, one on each side, so the cables form a closed loop. A ring network passes data from node to node around that loop until it reaches the device it is addressed to. Each node receives the signal and sends it on, so the ring needs no central hub.

The first ring networks were token-passing LANs. In IBM Token Ring, standardised as IEEE 802.5, a short frame called a token circulated around the ring and only the station holding the token could transmit, so there were no collisions. FDDI, an ANSI standard for 100 Mb/s over fibre, used two counter-rotating rings so that a single cut could be wrapped around. Both have been replaced by switched Ethernet in offices.

The ring layout survives in industrial networks for a different reason: redundancy. A cable run that loops through every cabinet and back to the start gives each switch two paths to everywhere else, so one cut fibre or failed switch does not split the network.

A ring sits between a star and a mesh. It uses far less cable than a full mesh but still gives every node a second path.

Ethernet cannot run on an active loop. A broadcast frame would circle the ring for ever and multiply, a broadcast storm that saturates every link within seconds. An Ethernet ring therefore blocks one link logically while keeping it cabled, and unblocks it only when a fault appears.

In Ethernet Ring Protection Switching (ERPS), defined in ITU-T G.8032, one switch is the ring protection link (RPL) owner and holds its port on the RPL blocked. When a switch detects a failed link it sends R-APS messages around the ring, the owner unblocks the RPL and traffic flows the other way. G.8032 targets switching in under 50 ms for rings within its design limits. The older answer, the Spanning Tree Protocol in IEEE 802.1D and its rapid version RSTP (now in IEEE 802.1Q), works on any layout but converges more slowly.

Most of our managed industrial switches support G.8032 ERPS with recovery in under 50 ms, alongside STP, RSTP and MSTP for other layouts.

ERPS is the ring protocol most managed industrial switches now share, because it is an open ITU-T recommendation rather than one vendor’s scheme, so switches from different makers can sit in the same ring.

When a link fails, the two switches next to it block their ports onto the failed link, flush their MAC address tables and send signal-fail messages both ways around the ring. The RPL owner then unblocks the protection link and traffic flows the other way. G.8032 sets a target of under 50 ms for this switch-over on a ring of up to 16 switches and 1,200 km of fibre, which is why datasheet recovery figures usually come with a maximum ring size.

Three timers shape what happens around the switch-over. A hold-off timer, zero by default, gives a lower-layer protection scheme time to act first. A guard timer, 500 ms by default, stops outdated messages still circling the ring from undoing a recovery. In revertive mode, a wait-to-restore timer, 5 minutes by default, keeps traffic on the backup path until the repaired link has stayed up long enough to trust; non-revertive mode leaves traffic where it is until an operator moves it.

Each ring is built from a few defined roles:

Recovery time is the figure that matters on a control network, because a controller that loses its I/O for longer than its watchdog time trips the process.

Recovery times depend on ring size, link speed and the switch, so take the figure from the datasheet. Many switch vendors also offer proprietary ring protocols. They can be fast, but every switch in the ring then has to come from the same vendor.

Rings are common wherever the equipment is spread along a line and an outage is expensive.

Plan the ring around the longest link and the harshest cabinet. Fibre suits long or electrically noisy runs between buildings, and our guide to choosing single-mode or multimode fibre covers the distance limits. Copper suits short links inside a building.

Use the same ring protocol on every switch, keep the ring within the size the protocol is specified for, and decide which switch owns the blocked link. On rolling stock, M12 Ethernet switches with locking connectors suit vibration, and some include bypass protection that keeps a daisy chain passing traffic if one switch loses power. Our guide to choosing managed or unmanaged industrial switches explains where a ring is worth the extra configuration.

We supply managed industrial switches for Ethernet rings and set up the ring protocol, VLANs and IP addressing before they ship, so the ring is ready to connect on site.