On CBR Service
|
Abstract:
Although CBR service is relatively well understood, there are several open issues that must be resolved before large scale CBR networks can be provisioned and built. In this paper, we investigate the performance of CBR traffic in the context of large-scale networks, where many connections and switches coexist and interact. For this, we develop a framework for simulating such networks, decoupling the influence of breadth and depth. The performance metrics used are the end-to-end delay histogram and metrics derived thereof, such as the 99%-percentile, the mean or the histogram width, and the index of dispersion for intervals (IDI), which we use to assess bunching (or clustering) in superposition streams. Our results are briefly as follows: we found that CBR traffic can be efficiently transported by the First Come First Served (FCFS) scheduling discipline, which has the least implementation cost. Delays incurred by cross traffic composed of many CBR streams with different bandwidths and phases do not exceed a few cell times even under heavy load, which means that buildout buffers of 10 to 20 cells seem to be sufficient after traversing 20 switches. We also show that the Round Robin (RR) and Weighted Round Robin (WRR) disciplines are ill suited for CBR traffic, both in terms of performance and implementation complexity. We compare two analytical approximation methods, based respectively on M/D/1 queues and on the Multiclass Parametric Decomposition Method, with the simulation results and found them to be suitable to estimate delays for the FCFS discipline. 1
