Performance analysis of stochastic multi-server systems

As multi-server systems have become increasingly common in various kinds of networks and network applications, developing a methodology to analyze the performances of multi-server systems is very needed. Stochastic network calculus is a great theory for performance guarantee analysis. However, very few analytical results based on stochastic network calculus are available for multi-server systems, especially for stochastic servers. By the parameter named merge availability and some properties we proposed in this paper, a stochastic multi-server system can be simplified to a stochastic single-server system. On this basis we analyze the performance bounds of stochastic multi-server systems with some different schedulers.