Providing QoS Guarantees in Packet Switches

The continuous growth that we are observing in the demand for diversified Quality-of-Service (Qos) guarantees in broadband networks introduces new challenges in the design of next-generation packet switches. Packet scheduling is the most critical function involved in the provision of individual QoS guarantees to the switched flows. Most of the scheduling techniques proposed so far assume the presence in the switch of a single contention point, residing in front of the outgoing links. Such an assumption collides with the current trend in the development of switches whose total cost scales well with the aggregate switched bandwidth. In these scalable switches, buffers and basic functionalities are distributed over multiple modules, and mainly located at the port cards that control the external inks. Buffering packets at the ingress port cards implies the introduction of an additional contention point in the system, and requires the definition of a distributed scheduling architecture to retain the QoS features of switches having completely centralized buffers. In this paper, we define a distributed scheduler that provides differentiated QoS guarantees to individual traffic flows with heterogeneous requirements and is suitable for application in QoS frameworks for both IP and ATM networks.