What is MRP (Media Redundancy Protocol)?
The Media Redundancy Protocol (MRP) is a network protocol defined in IEC 62439-2 that lets an industrial Ethernet network run as a ring and, after a cable break or a failed switch, automatically activates an alternative path within a guaranteed time, typically 200 milliseconds. MRP was developed in the PROFINET world, where it is the standard for network-level redundancy, but it also works with other Ethernet protocols. It makes ring redundancy predictable, which general-purpose IT protocols cannot achieve.
🧠 How does MRP work?
In an MRP ring every device has two ring ports. One device is the Media Redundancy Manager (MRM); all others are Media Redundancy Clients (MRC).
- The MRM logically blocks one of its two ring ports, so that no loop exists
- The MRM continuously sends test frames around the ring through both ports
- If the test frames stop returning, the ring is broken
- The MRM unblocks the blocked port and announces the topology change to all clients
- The switches flush their address tables and traffic flows via the other side of the ring
A device with the MRA role (Media Redundancy Automanager) automatically elects which device becomes the MRM, so no fixed manager has to be configured.
🔧 What recovery times does MRP offer?
| Profile | Maximum recovery time | Application |
|---|---|---|
| MRP 500 ms | 500 ms | Less time-critical installations |
| MRP 200 ms | 200 ms | Default; up to 50 devices in the ring |
| MRP 30 ms | 30 ms | Faster applications, smaller rings |
| MRP 10 ms | 10 ms | Highly time-critical applications |
With 200 ms recovery, a PROFINET installation usually does not notice the interruption, because the O watchdog time is often longer. For controllers that must not see any interruption at all, there is MRPD (Media Redundancy for Planned Duplication): with IRT, frames are sent around both sides of the ring simultaneously, making recovery seamless.
🔄 What is the difference between MRP, RSTP, DLR and PRP?
| Protocol | Standard / origin | Topology | Recovery time |
|---|---|---|---|
| MRP | IEC 62439-2, PROFINET | Ring | 10–500 ms, guaranteed |
| RSTP | IEEE 802.1w | Any topology | Usually sub-second, not guaranteed |
| DLR | ODVA, IP | Ring | Around 3 ms |
| PRP / HSR | IEC 62439-3 | Two parallel networks / ring | 0 ms (seamless) |
RSTP is more flexible, but its recovery time depends on network size and cannot be guaranteed. MRP trades that flexibility for predictability. PRP and HSR send every frame twice, which makes them more expensive, but without any interruption.
🏭 Where is MRP used?
- Production lines with PROFINET - the ring connects the PLC, O and drives
- Process industry - rings between the cabinets of PCS 7 and PCS neo installations
- Infrastructure - tunnels and water treatment plants where cabling spans long distances
Managed industrial switches from vendors such as Siemens (SCALANCE) and Phoenix Contact support MRP out of the box.
🔐 What should you watch out for with MRP?
- Exactly one MRM per ring - a misconfigured second manager causes disruptions
- No ring without MRP - a ring without an active redundancy protocol causes a broadcast storm that brings down the entire network
- MRP is not security - the protocol has no authentication; spoofed test frames can disrupt the ring. Protect the ring network with segmentation and port security
- Monitoring - a ring that silently falls back to the backup path is no longer redundant; have the MRM raise an alarm via SNMP or PLC diagnostics
🛠️ How do you set up an MRP ring?
- Choose the ring members - only devices with two MRP-capable ports, such as managed industrial switches or PROFINET devices with an integrated switch
- Assign the roles - one device as MRM (often a central switch or the PLC), the rest as MRC; or use MRA for automatic election
- Choose the recovery profile - usually 200 ms; check that the number of devices stays within the profile’s limit
- Configure the domain name and ring ports on all members, and make sure the ring is only physically closed once MRP is active
- Test the redundancy - pull a cable during commissioning and check that control continues without disruption
- Document and monitor - record the ring in the CMDB and have an open ring raise an alarm
| Check | Why |
|---|---|
| I/O watchdog time > MRP recovery time | Otherwise I/O drops out during switchover |
| At most one active MRM | Two managers cause instability |
| No RSTP on the same ring ports | Two redundancy protocols in one ring conflict |
❓ Frequently asked questions
What does MRP stand for in networking?
In industrial networks, MRP stands for Media Redundancy Protocol, standardised in IEC 62439-2. The protocol automatically restores an Ethernet ring network within a guaranteed time after a cable break or a failed switch. Note that in logistics MRP also stands for Material Requirements Planning, which is something entirely different.
How fast does MRP restore a network?
MRP restores a ring network within the maximum time of the selected profile: 10, 30, 200 or 500 milliseconds. The most commonly used profile is 200 ms with up to 50 devices in the ring. For seamless redundancy without any interruption there is MRPD, in combination with PROFINET IRT.
Is MRP only for PROFINET?
MRP comes from the PROFINET world and is the standard for ring redundancy there, but it is an open IEC standard that works at Ethernet layer 2. Many industrial switches also support MRP for other traffic, such as EtherNet/IP or Modbus TCP. In EtherNet/IP environments, however, DLR is more commonly used.
Which is better: MRP or RSTP?
MRP is better when a guaranteed, short recovery time is needed in a ring topology, as with PROFINET controllers. RSTP is more flexible and works in any topology, but its recovery time depends on network size and is not guaranteed. In industrial rings MRP is therefore usually preferred.
📌 In summary
MRP is the PROFINET ring redundancy protocol (IEC 62439-2) that restores a ring network within a guaranteed time, 200 ms by default, after a cable break or a failed switch. Where RSTP is flexible but unpredictable, MRP delivers the deterministic recovery time that industrial control systems require.