Simulating Queueing Systems
Most realistics computer and communication systems are very complex networks of subsytems and the jobs processed by such systems involve wide-ranging service requirements. In many cases the arrival and service processes are non-Markovian and the service discipline may use priorities in scheduling service requests. While Mean Value Ananlysis can be used to obtain average performance values for a computer and communication system, the result will still be based on simplifying assumptions. Discrete event simulations can be used to model and analyze complex systems exhibiting intricate interactions. In this chapter, we describe how discrete event simulations can be designed to represent large computer sysytems. A detailed MATLAB based simulation of M/M/1 queuing network is included. Other open source simulation tools are also introduced.