Access Traffic Control Implementations for Frame Relay

摘要

This paper considers various aspects of the implementation of access traffic control mechanisms for Frame Relay networks.We highlight the distinction between two related rate control mechanisms:access policing by the network and traffic shaping by the user.Frame discarding performance (policer strictness)is compared for two alternative poficing techiques,leaky bucket and quantized moving window.Hardware and software implementations are also compared for these two policers.It is estimated that between 15 and 150 policers could be implemented on a single 10 mm by 10 mm chip with frame throughputs per policer of a few million frames per second.Processor sharing would be possible for large-scale systems,with correspondingly lower frame throughputs.We also discuss design requirements of a leaky bucket shaper that will achieve zero policing loss and minimum delay when operating in conjunction with a leaky bucket policer.The loss/delay tradeoff for the shaping function is also quantified.