RIP: Why is spilt horizon with poisson reverse better?

Hi,

I have a fair understanding of Spilt Horizon and Spilt Horizon with poisson reverse but I am unable to get a concrete answer on why spilt horizon with poisson reverse is better than spilt horizon as infact in spilt horizon, all those route entries which were originally learnt from a network are omitted before sending an RIP update to that network, so it reduces the size of RIP Response packets and thereby also reduce the time involved in processing these Responses at the receving router. I have read that poissson reverse provides more robustness, but it isn't exactly clear to me!!

Cheers,

Datacom_guy

If you go to firewall.cx home page then Networking - Routing - Routing Protocols - Distance Vector. Then scroll toward the end of the article and you will find an example of how loops could be formed then a discussion of Maximum Life (TTL), Split horizon, Route Poisoning (poison reverse), and Hold down timers. These all help to prevent routing loops within Distance Vector routing protocols.

One thing not discussed though is Triggered updates. "A triggered update is sent immediately in response to some change in the routing table. The router that detects a topology change immediately sends an update message to adjacent routers. These routers generate triggered updates to notify their adjacent neighbors of the change. When a route fails, an update is sent immediately. Triggered updates, used in conjunction with route poisoning, ensure that all routers know of failed routes before any holddown timers can expire.

Triggered updates do not wait for update timers to expire. They are sent when routing information has changed. A router will immediately send a routing update on its other interfaces. This forwards the information about the route that has changed and starts the holddown timers sooner on the neighbor routers. The wave of updates propagates throughout the network." (This is a quote from the new version 3.1 Cisco Academy curriculum. Keep in mind if you don't give credit to quotes, especially copyrighted material, it's a triggered update is sent immediately in response to some change in the routing table. The router that detects a topology change immediately sends an update message to adjacent routers. These routers generate triggered updates to notify their adjacent neighbors of the change. When a route fails, an update is sent immediately. Triggered updates, used in conjunction with route poisoning, ensure that all routers know of failed routes before any holddown timers can expire.

Triggered updates do not wait for update timers to expire. They are sent when routing information has changed. A router will immediately send a routing update on its other interfaces. This forwards the information about the route that has changed and starts the holddown timers sooner on the neighbor routers. The wave of updates propagates throughout the network." (This is a quote from the new Version 3.1 Cisco Academy curriculum. One should always give credit to copyrighted material. Failure to do so is Plagiarism.)

This helps speed up convergence with Distant Vector routing protocols.

Good to see you around after a long time Mew.

Cheers,

Thanks mew for a clear description about Triggered Updates. But I still don't understand whether Spilt Horizon with poisson reverse is better than Spilt Horizon and if yes, why? I went through Firewall.cx link but couldn't have a clear understanding.

Cheers,

Datacom_guy

Split horizon by itself does not address all the possibilities when it comes to preventing routing loops. I realize I did not directly answer your question on why poison reverse makes split horizon more robust. I was trying to make the point that adding all the techniques used by distance vector routing protocols are needed to address the many situations that might cause a routing loop. No one technique by itself will do the job. As you add all of them together they do become more (“robust”?) efficient at preventing routing loops.

Understanding the need for each of the techniques and how they complement each other is what is important. On the other hand to answer your question directly, as best I can without a diagram to illustrate where this would apply, you should note that poison reverse breaks the split horizon rule in that its rule is “Once you learn of a route through an interface, advertise it is unreachable back through that same interface”.

If you have access to the CCNA 1 and 2 companion guide third edition refer to pages 717 and 718 and figure 16-7.

Also, note that poison reverse is different from route poisoning in that route poisoning does not break split horizon rule.