System Modeling and Performance Analysis of Cloud VirtualMachine Allocation Strategy with Entry Control and Reservation Mechanism

With the rapid development of the Internet, the number of requests in cloud computing is increasing rapidly. However, cloud resources are limited, and the surge in demand will lead to performance degradation in cloud services. To improve the performance of cloud services, we propose a cloud virtual machine allocation strategy based on reservation mechanism and entry control. In the presented strategy, a dynamic probability threshold is adopted to suppress the interference of low-priority requests on high-priority requests, while reserved virtual machines and cache resources are deployed to guarantee the service quality of low-priority requests. Based on queueing theory, we construct a mathematical model of the proposed strategy as a multi-server discrete-time queueing model. A three-dimensional Markov chain is constructed by abstracting system states to describe the stochastic evolution of the cloud service system. After solving the steady-state probability distribution, analytical expressions of system performance metrics are derived. Numerical experiments are conducted to analyze the variation trends of performance indicators, and comparisons with conventional strategies validate the effectiveness of the proposed strategy. Finally, a system benefit function is established to determine the optimal number of reserved virtual machines for maximum system profit.