PoisonReverse

PoisonReverse is a loop-avoidance technique used in some distance-vector routing protocols to prevent routing loops. It extends the idea of split horizon by not only avoiding sending information about a route back on the same interface, but by explicitly advertising that the route is unreachable to the neighbor from which it was learned.

How it works: When a router A learns a route to a destination D via a neighbor B, A will send updates to B in which the metric for D is set to an infinite value (poisoned). This tells B that D cannot be reached through A. By advertising an infinite metric back to the source of the learned route, the neighbor is discouraged from using this router as the next hop for that destination.

Relation to split horizon: PoisonReverse is often described as split horizon with poison reverse. Split horizon prevents advertising a route back on the same interface, and poison reverse takes this further by explicitly signaling that the route is unreachable through that neighbor.

Example: If A learns a path to D from B, A will advertise to B that the distance to D is infinity. B, seeing D as unreachable through A, will avoid using A to reach D and will look for alternative paths it may have learned from other neighbors.

Limitations: PoisonReverse does not guarantee loop-free operation in all topologies, especially in complex networks with multiple interconnections. It also increases routing update traffic and may not be sufficient alone in modern networks, where other loop-avoidance mechanisms are used.

See also: split horizon, distance-vector routing, RIP.