Optimal Multiserver Configuration for Profit Maximization in Cloud Computing

Simulations have been conducted for two purposes, namely, (1) to validate our analytical results (Theorems 1 and 2); (2) and to find more effective queueing disciplines which increase the net profit of a service provider. In Table 1, we show our simulation results by using the same parameters in Figures 1–10. For each λ = 6.05, 6.15, ..., 6.95, we trace the behavior of an M/M/m queueing system with the FCFS queueing discipline by generating a Poisson stream of service requests with arrival rate λ, recording the waiting and response times of each service request, and calculating the service charge to each service request. The average service charge of 1,000,000 service requests is reported in Table 1 for each λ. Notice that the maximum 99% confidence interval of all the data in the table is ±0.5165372%. The analytical data in the table are obtained by Theorem 2 to calculate the expected charge to a service request. It is easily observed that our simulation results match with the analytical data very well. These results validate our theoretically predicted service charge in Theorem 2, which is based on our analytical result on waiting time distribution in Theorem 1. Our analysis in this paper is based on the FCFS