A ring that heals a cut cable in a blink is beautiful engineering. A ring assembled by vibes is a broadcast storm scheduled for a random Tuesday. The difference is four rules.
The rules
- Exactly one MRP manager per ring (a capable switch or CPU port pair) with all other participants as clients; the manager blocks one ring port in normal operation — the loop is logically open, which is why it works at all. Two managers, or none, are both storm recipes.
- The ring is closed in configuration before it is closed in copper. Patch the final link only after MRP roles are downloaded — the temporary unmanaged loop is the classic commissioning outage.
- Watchdogs sized for reconfiguration: healing takes up to the 200 ms class (protocol worst case); Profinet device watchdogs shorter than the healing time convert a survivable cable fault into a device failure festival. Audit every device’s update × retry math against it.
- The ring is documented: which ports are ring ports on every participant, wall-mounted. Maintenance plugging a spare into a ring port creates topologies MRP never promised to survive.
Test it or you don’t have it
At commissioning: pull each ring segment live, confirm production continues, confirm the alarm arrives (a healed fault without an alarm is a hidden single point again — MRP masks faults from production, never from maintenance), replace, confirm re-closing. Ten minutes per segment, and the plant actually owns what it paid for.
FAQ
MRP or plain line for a machine? Rings pay where a cable fault costs more than the second cable path — long conveyors, cranes with festoons, multi-panel lines. A one-panel machine rarely needs one.
Faster healing options? MRPD / IRT-class mechanisms serve motion domains; standard MRP serves the general case — matching the tier to the need is the design act.
Zone Otomasyon commissions rings by cutting them — on purpose, with a stopwatch. Networks that survive.