Cases when Poison Reverse is used

Unanswered Question
Mar 5th, 2009

When is Poison Reverse used?

By default I thought in Cisco routers the implemented split horizon is with poison reverse.

So I expected to see that after R2 receives from R1 a route with a certain metric M1, the R2 would advertise back to R1 the same route but with metric 16.

However what I observed is different:

- either the router R2 didn't advertise back the route at all

- or the router R2 advertised back to R1 the route but with the same metric M1 (this latter case was when R1 received that route from another router R3, but with metric M1-1)

Do I miss something?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (1 ratings)
Loading.
Giuseppe Larosa Thu, 03/05/2009 - 13:52

Hello Petru,

split horizon is not enough is some redundant topologies.

possible definitions are :

route poisoning:

The process of sending an infinite-metric (hop count 16) route in routing updates when

that route fails.

poison reverse:

The act of advertising a poisoned route (metric 16) out an interface, but in reaction to

receiving that same poisoned route in an update received on that same interface.

What you have seen is:

split-horizon

and a better route installed that is proposed on the same common subnet:

a third router could install both routes with the same hop count.

To see poison reverse in action I would:

have a connected network advertised by R1 to R2.

you shut that interface

R1 should send out a triggered update with hop count 16

R2 should react sending out an update with metric 16

If I remember correctly this is the test that Edison did in another thread about poison reverse.

Hope to help

Giuseppe

badalam_nt Fri, 03/06/2009 - 12:47

So the 2 cases I mentioned are correct, but additionally there are 2 other ones:

3) when R2 receives from R1 the route with metric 16, but it receives as well the same route from another router R3 with a metric M3 different from 16, the router R2 will advertise back to R1 the route with metric M3+1.

4) when R2 receives from R1 the route with metric 16, and it doesn't receive the same route with a metric strictly lower than 15 from any another router, the router R2 will advertise back to R1 the route with metric 16.

And only this last case 4) is the one associated to poison reverse.

Is my summary correct?

Giuseppe Larosa Fri, 03/06/2009 - 13:12

Hello Petru,

I agree:

poison reverse is an explicit update of non-reachability that overrides the split-horizon rule:

the router sends a message like: "don't consider me as a possible alternate path to reach that route: you cannot reach it and also I cannot"

Hope to help

Giuseppe

Actions

This Discussion