Recent market demand has put quality of service (QoS) support as the key feature in differentiating network products from various vendors. Advances in silicon technology have made it feasible to design per-flow-based traffic management algorithms to control QoS with substantially improved performance over the traditional class-based approach for next generation switches and routers. This paper presents a node architecture with the aim of achieving QoS provisioning for guaranteed service (GS), controlled-load (CL), and best-effort (BE) service offerings for future integrated services networks. Under our node architecture, we propose several novel traffic management algorithms, which include Adaptive Rate allocation for Controlled-load (ARC) flows, a hybrid model-based and measurement-based admission control algorithm for GS and CL flows, and a Quasi-Pushout Plus (or QPO+) packet discarding mechanism. Simulation results show that, once admitted into the network, our node architecture and traffic management algorithms are capable of providing hard performance guarantees to GS flows under all conditions, consistent (or soft) performance to CL flows under both light load and heavy load conditions, and minimal negative impact to conforming GS, CL, and BE traffic should there be any non-conforming behaviour from some CL flows. Furthermore, our node architecture and traffic management algorithms resolve some key problems associated with the traditional class-based approach.
|Original language||English (US)|
|Number of pages||18|
|Journal||International Journal of Parallel and Distributed Systems and Networks|
|State||Published - Dec 1 2000|
ASJC Scopus subject areas
- Hardware and Architecture