Consider the below Scenario,
R3 (Network 10.0.0.0)
R1 is connected to both R2 and R3 R1, R2 and R3 are all connected in a traingle topology , R3 has Network (10.0.0.0) on its local LAN.
Now, R1 knows how to reach Network 10.0.0.0 via the direct link to R3 and the link through R2, when R3 Looses its connection to Network 10.0.0.0 (For example the interface goes down), R3 advertises the Network 10.0.0.0 with Infinity metric (Metric 16) to R2 and R1.
R2 then advertises the same Network with Infinity metric to R1. R2 poison the route to R3 for Network 10.0.0.0 as well as R1, and the Hold down timer starts on R2,R1m Respectively, after the Hold down timer expires and there is no update from R3 if the Network is reachable or not, R2 waits for the flush timer to expire and immediately removes the Path to Network 10.0.0.0 from its routing table.
The Poisoning here help in avoiding the loop because if Router 2 wouldnt poison the route to Network 10.0.0.0, R1 would still think the possible path to Network 10.0.0.0 is through R2
and If R1 wouldnt poison the route to R3 for Network 10.0.0.0, it will still think the Network is reachable via R2 and thus creating routing loops.
So, a loop can occur in a fully meshed RIP Network if we dont have The Poisoning mechanism.
Notes, The Update in Rip is periodic update and takes place every 30.
Distance vector protocol keeps track of any changes in network and broadcast periodic updates on all active interfaces so suppose when the link is down inorder to avoid loops the router initiates route poisoning by advertisiing that network as 16 or unreachable,which in turn it receives it as poison reverse....so when a router receives a route poison it deletes the route information from the routing table..
Ex:How loop occurs?
lets consider the 4 routers are connected, by routing updates , all routes know path to every route...suppose if link between router3 and router4 fails,router3 tells router2 to stop updating to router4 but router1 dont know that information yet, for it the link is availiable with a metric 3 so it sends updates ,router2 receives this update and thinks that router4 is availiable through router1 so it updates it routing table with a metric 4,this process continues and loop forms..
route poisoning is used with triggered updates..
thanks all ! Thanks Keith! the vedio was informative. I proceeded to read further more and realised there are triggers which are sent as as well one way to avoid loops ,
The triggered updates are primarily to assist the network in converging faster. Poison reverse and split horizon are primarily to avoid loops, and you are right, the poisoned triggered updates work to avoid loops.
Thanks for the feedback,
Durga is doing a great job in our study group. I wanted to also add that route poisoning is used by various DV routing protocols in order to overcome large routing loops and offer explicit information about when a subnet or network is not accessible or merely suggesting that the network is unreachable by not including in its updates. Typically this is accomplished by setting the hop count to one more than maximum. This could be directly associated with poison reverse updates. These updates messages are transmitted by a router back to the originator (thus ignoring the split-horizon rule) after router poisoning has occurred.