Burst and fast packet switching: Performance comparisons

Abstract

Abstract Two alternatives to the prevailing view of the ISDN backbone are burst switching and fast packet switching. Both these technologies have the potential to provide considerable savings in transmission and switching resources over systems with less integration, and treat the traffic generated by different services in a unified manner. In this paper we explore the significant differences between them from the performance perspective. Other important issues for a complete comparison of the two technologies, such as evolutionary development, processing complexity, and variable bit-rate encoding, are not considered. Burst and fast packet are first compared from a voice-only perspective. For the same TASI advantage and equal capacity, it is shown that fast packet can provide the same performance as burst switching. However, under these circumstances, the average residual capacity left for data is significantly greater for burst than fast packet. Nevertheless, using analytic models for both systems, it is shown that data performance within that residual capacity is — for 64 kb/s encoded voice — not significantly different except for regions of high utilization. A model developed and validated in earlier work is used to analyze the data performance in burst switching. To analyze the performance of data on an integrated packet-switched link, an analytic model is developed based on an earlier renewal theory model for packet voice and using generalized results for priority queues.