Study of Performance Measures and Energy Consumption for Cloud Computing Centers Based on Queueing Theory

Abstract

Cloud computing, as an effective method to provide computing resource and service to users on demand, is attracting more and more attention. This paper establishes a GI/G/m queue to characterize the cloud computing centers with general arrival process and service times. We analyze the main performance measures such as the mean waiting queue length and average waiting time of service requests in the cloud computing system. Besides, the energy consumption model of the cloud computing center is established. The results show that increasing the service rate can reduce the average waiting time, the quantity of running servers needs to be varied with the arrival rates of service requests to meet the satisfaction of users. The total energy consumption is connected with the arrival rate, service rate and the quantity of servers in the cloud computing center, but it has nothing to do with the coefficient of variation squared of the interarrival times and service times.