The impact of count to infinity in distance vectors?
High latency due to unneeded packet loss for unreachable networks.
What have poison reverse solved?
In distance vector, split horizon prohibits route advertisements sent to interfaces where routes were originally learned. While split horizon avoids loops, in the case of an unreachable network the dynamic vector routing tries to find alternative paths from adjacent neighbors.
R1, R2 and R3 are directly connected. R3 is advertising loopback0 in EIGRP. R1 and R2 learns network from R3 and both sends topology changes to each other. As we have split horizon enabled on R3’s interface, it never receive it’s own routes. The flow is manipulated so that traffic destined from R3 to R2 is preferred through R1.
In an activity where we lose complete connectivity to R3, EIGRP would scan through the topology table and observer that R1 still has a valid route to R3.
Due to the lack of re-transmission of outdated information and slow convergence in distance vector algorithms, split horizon incorporated poison reverse by adding a cost of 16 (know as Infinity). The count to infinity is able to converge information in a few seconds.
You can simulate the poison reverse on GNS3 and observe EIGRP query packets.
After the 3.3.3.3 network goes down, split horizon is initiated on the physical link facing the N originator. Reply messages are sent to adjacent destinations and all records are flushed.
Leave a Reply