Split horizon explained
Distance vector protocols are susceptible to routing loops. Routing loops occur when a packet is continually routed through the same routers over and over, in an endless circle. Because they can render a network unusable, distance vector routing protocols (such as RIP and EIGRP) employ several different mechanisms to prevent routing loops.
Split horizon is one of the methods used by distance vector routing protocols to avoid routing loops. The principle is simple – a router will not advertise a route back onto the interface from which it was learned. Split horizon is enabled on interfaces by default.
To consider what could happen without the split horizon mechanism, take a look at the following example:
We have a network of three routers. All routers are running RIP, a distance vector protocol. R3 is directly connected to the 10.0.0.0/24 network and advertises that network using RIP to R2. R2 receives the routing update, places the route in its routing table and informs R1 about the 10.0.0.0/24. Because the split horizon mechanism is enabled by default on all interfaces, R1 will not advertise to R2 that it has the route to 10.0.0.0/24.
Now consider what would happen if the split horizon mechanism didn’t prevent R1 to advertise the route back to R2. R1 would advertise to R2 that it has a route to reach the 10.0.0.0/24. Let’s say that the link between the R2 and R3 fails. Since R2 received a route to that network from R1, it will send all packets destined for the 10.0.0.0/24 network to R1. But R1 thinks that R2 has a route to reach that network (it doesn’t know that the link between R2 and R3 failed) and sends the packets back to R2, thereby creating a routing loop.