TABLE OF CONTENTS In Chapter 1, we defined the key service level agreement (SLA) parameters for IP applications and services as being: delay, delay variation or jitter, packet loss rate, throughput, availability, and per flow sequence preservation. In Chapter 2, we described the QOS component mechanisms and architectures that can be used in engineering a network to meet SLA targets. In this chapter, we build on the context and understanding created by Chapters 1 and 2, to show how the Differentiated Services architecture (Diffserv) can be practically deployed at the network edge and in the network core in order to satisfy defined application SLA requirements. Diffserv is by far the most widely deployed IP QOS architecture; it is widely deployed both in private enterprise networks and in service provider (SP) networks providing virtual private network (VPN) services to enterprises.
The edge of the Diffserv domain represents the provider/customer boundary for the Diffserv-enabled services being offered; Diffserv achieves scalability by performing complex per-customer QOS functions and maintaining per-customer state, only at the edges of the network. Hence, the policies applied at the edge of the network are different from, and generally more complex than, those applied within the core. Many networks are built with a hierarchy consisting of core routers (CRs), which provide connectivity between distribution routers (DRs), which in turn aggregate connections to routers at remote sites, each of which have local access routers (ARs). If, for example, we consider a Diffserv deployment in this type of network, as shown in...
