Route selection based on network topology without taking into account the information on the current network load was the main operating principle of routing protocols in packet-switched networks.

For each source address-destination address pair, such protocols chose the only route without accounting for information flows traveling through the network. As a result, all flows between these pairs of end nodes passed along this route, the shortest route according to metric measurements.

The chosen route may be either more rational (e.g., accounting for the nominal bandwidth of communications or delays introduced by these links) or less rational (e.g, accounting for only the number of transit routers be tween the source and the destination nodes).

The fish, the network with the topology shown in below figure , is a classic example illustrating the inefficiency of this approach. Despite the two routes between the A and E switches the upper one via the B switch and the lower one via the C and D switches the entire traffic from switch A to switch E will be forwarded along the upper route according to traditional routing principles. This happens for one reason, namely, that the lower path is longer by one transit node. Therefore, this route is ignored, although it could work in parallel with the upper one.

As a result, even the shortest route is congested, and the packets will be sent along this path anyway.