route poisoning

May 21, 2005 2 Replies

Rate this thread:

vertigo 21 years ago

Hello

What is route poisoning used for ? I wrote that it's needed to eliminate loops.

But when i use Split horizon i do not have loops already, so why do i need route poisoning ?

How exactly does it work, i my CCNA materials it's not written exactly what is happening. Example: RB----RC----RE(networkE)

I use RIP. When networkE is down routerE sends to routerC info about networkE with metrics 16. And after that RouterC will send to RouterB similar information about networkE (every 30 s)? (even when it received from RouterB info about networkE with metric 3) ? (I assume that i don't use Split horizon in this example)

Could anybody clarify this ?

So: what is the difference between using and not using route poisoning. I know that when we use routing poisoning we send data with metrics=max+1, but what are the implications - what is that for ? If we do not use route poisoning and the network is down we will not send info about that network and the result should be the same !

Thanx Michal

Vote

Mark Scott 21 years ago

Route poisoning is used to prevent loops when you run DV protocols like RIP. instead of the route being removed from the routing table, it is advertised using an infinite metric of 16. This advertisement forces the route out of other router's tables because a route of Metric 16 is considered unreachable.

If poisoning was not used then the route would be advertised via a different router which would then propogate invalid information around the network.

Split Horizon says that an advertised route must not be sent back out of the interface it was learnt on. Yes, this does prevent loops but there are situations where you NEED to have advertisements sent back out of the same interface ie VLANs.

Also, Split Horizon wont cure all loops on its own especially if there are a number of routers in the network because they could still transfer bad information if a link goes down.

Regards

Mark

Vote

vertigo 21 years ago

RB----RC----RE(networkE) So if networkE goes down without route poisoning and without split horizon i would have on routerC tables: interface_to_RB networkE metrics=3 interface_to_RE networkE metrics=unrechable and in such case routerC would send all data to networkE thru RouterB.

If we have route poisoning in this situation routerC would have only: interface_to_RE networkE metrics=unrechable and he would return appriopriate ICMP message for packets destined to networkE ? Am i correct ?

So, In other words when router receives poisoned information about network he assumes that this network is not available on any interface at all?

I still does not what is the difference between:

If we advertise from routerA information about two networks to routerB, he knows about theese two networks, and then we advertise from routerA information about only one network to routerB, routerB deletes information about that second network (after some time) - so why bother and send information about that network with metrics = 16 if we know that if we will not send any info it will be removed?

When we use route poisoning even when routerC receives from routerE poisoned information, a second later he could receive from routerB old information about networkE (that's it's available 3 hops from him). What would the routerC do if he receives such information ? Discards it ? If yes what is happening when networkE goes up again ?

Thanx Michal

Vote