OVERVIEW

In the preceding chapters we described the trends in packet- and circuit-switched networks that, together, culminate in the design of ATM networks and in improvements of the Internet protocols. We saw in Chapter 6 that a network that combines an ATM transport layer, or an IP layer with some form of quality of service or class of service, with a high-speed physical layer such as SONET can potentially provide the large range of quality of service (QoS) necessary to support most applications. However, in order to provide this range of QoS,the network's resources (bandwidth and buffers) must be properly managed or controlled.

In this chapter you will learn the concepts and fundamental techniques used to control circuit-switched, datagram, and ATM networks in order to achieve efficient use of network resources. You will understand how different control techniques affect different network performance measures, and you will acquire the skills to evaluate those performance measures, including blocking probability, delay, and loss. You will learn the deterministic proposals of the ATM Forum for admitting new connections and for allocating resources to those connections. You will see that those proposals are based on worst-case scenarios and that more sophisticated proposals based on effective bandwidth can realize the gains from statistical multiplexing. From Chapter 2 you already know the requirements imposed by various applications; now you will be able to calculate how well a specific application can be supported by the bandwidth and buffers allocated by the network for that application.

In section 8.1...