Skip to Main Content
The router buffer sizing problem has been identified recently as an important problem in networking research. Contrary to static proposals and rules of thumb recent efforts have been attempting to develop simple practical algorithms for dynamic queue size management, where the problem is defined on a heuristic optimization basis and the objective is to track a volatile optimum point by continuously tuning at runtime the buffer queue size, thereby adapting to the current traffic conditions. On the basis of a robust extremum control solution to the problem, proposed recently, in this paper we evaluate the QoS improvements that can be obtained with a dynamic router buffer sizing strategy compared to static proposals. The analysis is based on a plethora of data collected from several representative Ns-2 simulation experiments. Simple SLAs are adopted to reflect the QoS requirements of different applications in generic Internet traffic scenarios.