Performance Measurement and Queueing Analysis with Low Blocking Probability of Serial Connection Networks

In this paper we propose the performance measurement and queueing analysis for the low blocking probability of serial connection networks with which we can estimate the system response time of serial Web servers. First, we calculate the system response time of the serial connection network. Second, we simulate the queueing model of the serial connection network and find the system response time. Third, we measure the system response time by using different numbers of ASP multiplication loops to represent the different service rates of the serial connection network. Fourth, we compare the simulation and modeling results with a low blocking probability, and we find that the maximum error is about 4.3158%. Fifth, we compare the simulation and measurement results with a low blocking probability of the serial connection from the different service rates of the servers. Sixth, we compare the simulation and the 1K, 10K and 100 K ASP multiplication loops with the system response time measurement; this displays a discrepancy of about 1.97%, 5.94% and 20.70% respectively. Hence we learn that the system response time of a serial connection network has an approximate linear relation with respect to the number of stages of the queues when the utilization of the system or the blocking probability is low.