Service-Differentiated Connection Set-Up

Abstract

In designing multimedia applications over a multi-service network, we face the problem of supporting applications that have QoS requirements. Thanks to the fruitful research efforts in high-speed networking and multimedia applications, there has been, however, significant progress in understanding of how to deliver a data throughput (i.e. packets) above a certain rate and within delay bounds. It is now common agreement that for real-time traffic, the delivery of the data packets in a correct order over a logical association between users is likely to follow the connectionoriented principle. For such a connection establishment, the wavelength-routing circuits are the natural candidates. In this context, given a set of constraints for an end-to-end connection, the question arises: what subset of constraints regarding the “end-to-end” QoS requirements will be taken into account for wavelength-routed connection accommodation, if any? The best QoS agreement is certainly end-to-end. However, this requires co-ordination across many network devices, which is a big network design challenge, expensive to build and difficult to manage (Figure 3-1). We will see later on that instead of this co-ordination across the multi-layer stacks, the QoS parameters are translated into the client network requirements for links, paths or signal quality constraints in the wavelength domain.