Network redundancy: RSTP, MRP, DLR, HSR and PRP
Comparison of industrial Ethernet redundancy methods by recovery time, topology and typical use — from RSTP and vendor rings to MRP, DLR and zero-loss HSR/PRP.
Industrial networks use ring or parallel topologies so that one broken cable or switch does not stop production. The methods differ in recovery time, topology and which devices must support them.
Comparison
| Method | Standard | Topology | Typical recovery | Notes |
|---|---|---|---|---|
| RSTP | IEEE 802.1Q (802.1w) | Any mesh/ring | ~ hundreds of ms to seconds, depends on topology | Vendor-neutral, good for backbones; too slow for many I/O systems |
| MRP | IEC 62439-2 | Single ring | ≤ 200 ms (≤ 500 ms profile), faster profiles exist | Used with PROFINET; one ring manager, others clients |
| Vendor rings | Turbo Ring (Moxa), HIPER-Ring (Hirschmann), FRNT (Westermo)… | Ring / coupled rings | ~ 20–50 ms typical claims | Same vendor in the ring |
| DLR | ODVA (EtherNet/IP) | Ring of devices with embedded 2-port switches | ~ 3 ms for 50 nodes (ODVA figure) | Ring through the I/O devices themselves |
| HSR | IEC 62439-3 | Ring, frames sent both ways | 0 ms (no frame loss) | Every node needs HSR; halves bandwidth |
| PRP | IEC 62439-3 | Two independent parallel networks | 0 ms | Double infrastructure; common in substations (IEC 61850) |
Choosing
- Check what the I/O watchdog allows: PROFINET devices time out after the configured update time × watchdog factor; if ring recovery is longer, I/O goes to fault. MRP + appropriate watchdog, or PROFINET system redundancy (S2/R1) for critical cases.
- EtherNet/IP I/O connections time out after RPI × multiplier — DLR keeps recovery below that for device-level rings.
- For SCADA/server backbones RSTP or vendor rings are usually sufficient.
- Zero-loss requirements (protection, substation automation, motion) → HSR/PRP.
Commissioning tips
- Configure exactly one ring manager (MRP/DLR supervisor) per ring; set the others as clients.
- Do not mix RSTP and a ring protocol on the same ring ports.
- Test by pulling each ring cable while monitoring I/O status and diagnostic buffers; record the measured interruption.
- Monitor ring status (SNMP traps, PROFINET diagnostics, DLR status object) — an open ring that nobody noticed has no redundancy left.
Connections
Open in Knowledge Graph →Loading the map…
Sources
Updated 2026-10-04