Congestion Control

We address the issues of congestion in networks—when the offered traffic exceeds the available link or buffer capacity. This is an area with a long history of different solution approaches, ranging from advanced reservations of capacity (as offered by RSVP, for example) to sophisticated congestion detection and avoidance mechanisms, which have been developed over many decades for the Internet, particularly within TCP. There is also a broad range of approaches to managing the queuing resources in the network, ranging from the simple first-in, first-out to more complex schemes that control which packets are dropped (such as random early detection) and in what order packets are sent out of the queue (such as fair queuing). Finally, we look at how traffic engineering has evolved to mitigate congestion by better distribution of traffic over multiple paths.