In the EtherNet/IP world, ring redundancy is not something you bolt on with external switches; it is built into the devices themselves. Device-Level Ring is ODVA's answer to the same problem MRP solves in the Profinet world, but it takes a different route. It relies on field devices that already contain a small two-port switch, wires them into a ring, and lets one of them supervise that ring with fast beacon frames so a break is caught and rerouted in a very short time. This guide explains what a Device-Level Ring is, how the ring supervisor and its beacons detect and heal a fault, and why DLR, rather than spanning tree or MRP, is the redundancy scheme an EtherNet/IP monitoring integrator will meet.
Device-Level Ring (DLR) in one line: A Device-Level Ring, DLR, is the ring-redundancy protocol defined by ODVA for EtherNet/IP networks. It is built on devices that contain an embedded two-port switch, wired into a ring, with one device acting as the ring supervisor. The supervisor keeps the ring loop-free and sends rapid beacon frames in both directions; when a break stops the beacons, it reroutes traffic, typically recovering in well under 3 milliseconds. It is the native EtherNet/IP redundancy scheme, distinct from spanning tree and from Profinet's MRP.
The idea that distinguishes DLR is that the ring is made of the field devices themselves rather than of separate infrastructure switches. Many EtherNet/IP devices, from I/O blocks to drives, include a small embedded switch with two Ethernet ports, so a device can be daisy-chained by connecting one port to its upstream neighbour and the other to its downstream neighbour. When those daisy chains are closed into a loop, the devices form a ring, and DLR is the protocol that makes that ring safe and redundant without any external managed switch being required to build it.
As with any ring, an unmanaged loop would flood the network, so DLR designates one device as the ring supervisor, and that supervisor prevents the loop by not forwarding ordinary traffic across the ring on one of its ports under normal conditions. The other devices are ring nodes that pass traffic through their embedded switches. This arrangement keeps the wiring simple, a single ring threaded through the devices, while giving the network a redundant path that becomes active the instant the ring is broken.
Because the redundancy lives in the devices, DLR is economical for EtherNet/IP machines and skids where the devices already have two-port switches. There is no need to buy and mount separate ring switches; the ring is inherent in how the devices are cabled. That native quality is why DLR is so closely associated with EtherNet/IP specifically, and why an integrator working on an EtherNet/IP installation is likely to encounter DLR rather than the ring schemes native to other protocol families.
The ring supervisor does two jobs at once: it keeps the ring loop-free and it constantly checks that the ring is intact. To check integrity it emits beacon frames from both of its ring ports at a rapid rate, and under normal conditions those beacons travel around the healthy ring and are seen by the supervisor and the ring nodes. The high frequency of the beacons is deliberate: because they are sent so often, the loss of them is noticed almost immediately, which is the key to DLR's very fast reaction time.
When a cable is cut or a device in the ring fails, the beacons can no longer complete their path, and the supervisor detects the break by their absence. It responds by unblocking the port it had been holding closed, which turns the standby half of the ring into an active path and restores connectivity between all the surviving devices. Ring nodes on either side of the break also flush and relearn their forwarding as needed, so traffic quickly finds its way around the fault. The whole sequence, from break to restored ring, typically completes in well under 3 milliseconds for a ring of moderate size.
That sub-millisecond-to-few-millisecond recovery is dramatically faster than a general spanning-tree reconvergence and faster than a typical MRP ring, and it is fast enough that most EtherNet/IP connections can be kept alive across a single ring break without the controller dropping its I/O. The beacon mechanism is what buys this speed: by watching for the disappearance of very frequent frames rather than waiting on timers measured in tenths of a second, DLR can react to a fault almost as soon as it happens. The supervisor also reports the ring's state, so the fault does not just heal silently but is visible for maintenance.
An integrator arriving at an EtherNet/IP network needs to recognise which redundancy scheme is in play, because the answer shapes both the network's behaviour and what the monitoring layer should watch. On EtherNet/IP the native answer is DLR, and it is important to understand that DLR is not spanning tree and not MRP. Spanning tree, including its rapid variant, is a general Ethernet loop-prevention protocol whose recovery is far slower and which is used on plant IT backbones rather than on fast control rings. MRP is the ring standard native to the Profinet world. DLR is the ODVA-defined scheme purpose-built for EtherNet/IP device rings.
The distinction affects diagnostics. A DLR ring supervisor maintains information about the state of the ring, including whether it is intact or faulted and where a break has occurred, and this information is available through EtherNet/IP's object model. That means a monitoring system speaking EtherNet/IP can read the ring's health directly, presenting whether the ring is whole, whether it is currently running on its redundant path, and details that help locate a fault. This is richer and more specific than what a spanning-tree network typically exposes about its own topology.
For a cloud SCADA or remote monitoring project the practical takeaway is to treat the DLR ring as a monitored asset in its own right. Because the supervisor exposes ring status as EtherNet/IP data, an edge collector can read it and forward it upward, so the ring's condition becomes a set of tags that can be trended and alarmed alongside the process data. A ring that has quietly failed over to its backup path is still working but has spent its redundancy, and surfacing that state to operations means a single cable fault gets fixed before a second one takes the network down. Recognising DLR as the scheme in use is the first step to monitoring it properly.
A Device-Level Ring typically recovers from a single break in well under 3 milliseconds for a ring of moderate size, fast enough that most EtherNet/IP connections stay alive without the controller dropping its I/O. This speed comes from the ring supervisor sending very frequent beacon frames, so the loss of those beacons signals a break almost immediately rather than waiting on slower timers.
No. Both are ring-redundancy protocols, but DLR is ODVA's scheme for EtherNet/IP, built on devices with embedded two-port switches and a beacon-based ring supervisor that heals in under 3 milliseconds. MRP is the IEC 62439-2 standard native to Profinet, using a manager and test frames with a typical recovery around 200 milliseconds. They are different protocols tied to different ecosystems and are not interoperable.
The ring supervisor is the one device in a DLR ring that keeps the ring loop-free and monitors its integrity. It sends rapid beacon frames in both directions around the ring, and when those beacons stop completing their path it detects the break and reroutes traffic onto the redundant path. It also maintains and reports the ring's status, which monitoring systems can read through EtherNet/IP.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.