Performance measurement and queueing analysis of parallel connection servers with different dispatch probabilities

This paper is a description of our efforts to create a queueing model for parallel connection servers with different dispatch probabilities, so that we can estimate the system response time of parallel Web servers. First, we calculate the system response time of the parallel connection servers with different dispatch probabilities. Second, we simulate the queueing model of the parallel connection servers with different dispatch probabilities and find the system response time. Third, we measure the system response time of the parallel connection servers with different dispatch probabilities. Fourth, we compare these simulation and modeling results whilst the CPU loading is low, and we find that the maximum error is about 0.1%. Fifth, we compare the measurement with the modeling results, which shows that the difference is about 11.41%-0.01%.