Designing Large Scale Lans

Chapter 6. IP Dynamic Routing

The network needs to understand how to get packets through from one side to the other. This can be accomplished in several ways. In a simple network with only one or two routers, it is probably most efficient to configure this routing information into the routers manually. However, in a large or complex network, the routers need to learn and update routing information through the network automatically. This is particularly true for networks that offer multiple redundant paths for fault-tolerance purposes.

Dynamic routing protocols give the network a way of healing around link or equipment failures. This is because they can see all of the different paths through a network and pick the best one at any given moment. When one path becomes unusable, another is selected.

The routing of IP packets is always handled by means of a routing table. This table is basically just a list of destination networks and the next hop required to get to these destinations. It may also contain other supplemental information, such as an estimate of how much it costs to use that particular route, and it may contain several different options for directing traffic to some destinations.

This concept of the cost of a route is relatively open and vague. The cost could be a function of any number of variables, such as the number of hops, the net latency of the path, the minimum bandwidth along the path, as well as other less tangible factors. For example, it may be better to ...