It’s important to note that our distance vector routing table has been almost completely built on secondhand information. Any route that a router reports with a hop count greater than 1 is based upon what it has learned from another router. When Router B tells Router A that it can reach Network 5 in two hops or Network 6 in three, it is fully trusting the accuracy of the information it has received from Router D. If, as a child, l you ever played the telephone game (where each person in a line receives a whispered message and tries to convey it exactly to the next), you quickly realize that secondhand information is not always as accurate as it appears to be.
Figure 3.8 shows a pretty simple network layout. It consists of four logical networks separated by three routers. Once the point of convergence is reached, each router will have created a routing table, as shown in the diagram.
Figure 3.8: Given the diagrammed network, each router would construct its routing table.
Now, let’s assume that Router C dies a fiery death and drops offline. This will make Network 4 unreachable by all other network segments. Once Router B realizes that Router C is offline, it will review the RIP information it has received in the past, looking for an alternate route. This is where distance vector routing starts to break down. Because Router A has been advertising that it can get to Network 4 in three hops, Router B simply adds 1 to this value and assumes it can now reach Network 4 through Router A. Relying on secondhand information clearly causes problems: Router B cannot reach Network 4 through Router A, now that Router C is offline.
As you can see in Figure 3.9, Router B would now begin to advertise that it can now reach Network 4 in four hops. Remember that RIP frames do not identify how a router will get to a remote network, only that it can and how many hops it will take to get there. Without knowing how Router A plans to reach Network 4, Router B has no idea that Router A is basing its route information on the tables it originally received from Router B.
So Router A would receive a RIP update from Router B and realize that it has increased the hop count to Network 4 from two to four. Router A would then adjust its table accordingly and begin to advertise that it now takes five hops to reach Network 4. It would again RIP and Router B would again increase the hop count to Network 4 by one.
Figure 3.9: Router B incorrectly believes that it can now reach Network 4 through Router A and updates its tables accordingly.
Note This phenomenon is called count to infinity because both routers would continue to increase their hop counts forever. Because of this problem, distance vector routing limits the maximum hop count to 15. Any route that is 16 or more hops away is considered unreachable and is subsequently removed from the routing table. This allows our two routers to figure out in a reasonable amount of time that Network 4 can no longer be reached.
Reasonable is a subjective term, however. Remember that RIP updates are only sent out once or twice per minute. This means that it may be a minute or more before our routers buy a clue and realize that Network 4 is gone. With a technology that measures frame transmissions in the micro-second range, a minute or more is plenty of time to wreak havoc on communications. For example, let’s look at what is taking place on Network 2 while the routers are trying to converge.
Once Router C has dropped offline, Router B assumes that it has an alternative route to Network 4 through Router A. Any packets it receives are checked for errors and passed along to Router A. When Router A receives the frame, it performs an error check again. It then references its tables and realizes it needs to forward the frame to Router B in order to reach Network 4. Router B would again receive the frame and send it back to Router A. This is called a routing loop. Each router plays hot potato with the frame, assuming the other is responsible for its delivery and passing it back and forth. While our example describes only one frame, imagine the amount of bandwidth lost if there is a considerable amount of traffic destined for Network 4. With all these frames looping between the two routers, there would be very little bandwidth available on Network 2 for any other systems that may need to transmit information.
Fortunately, the network layer has a method for eliminating this problem, as well. As each router handles the frame, it is required to decrease a hop counter within the frame by 1. The hop counter is responsible for recording how many routers the information has crossed. As with RIP frames, this counter has a maximum value of 15. As the information is handled for the 16th time (the counter drops to 0) the router realizes that the information is undeliverable and simply drops the information.
While this 16-hop limitation is not a problem for the average corporate network, it can be a severe limitation in larger networks. For example, consider the vast size of the Internet. If RIP were used throughout the Internet, certain areas of the Internet could not reach many resources.