Task-Driven Virtual Machine Optimization Placement Model and Algorithm

In cloud data centers, determining how to balance the interests of the user and the cloud service provider is a challenging issue. In this study, a task-loading-oriented virtual machine (VM) optimization placement model and algorithm is proposed integrating consideration of both VM placement and the user's computing requirements. First, the VM placement is modeled as a multi-objective optimization problem to minimize the makespan of the loading tasks, user rental costs, and energy consumption of cloud data centers; then, an improved chaos-elite NSGA-III (CE-NSGAIII) algorithm is presented by casting the logistic mapping-based population initialization (LMPI) and the elite-guided algorithm in NSGA-III; finally, the presented CE-NSGAIII is employed to solve the aforementioned optimization model, and further, through combination of the above sub-algorithms, a CE-NSGAIII-based VM placement method is developed. The experiment results show that the Pareto solution set obtained using the CE-NSGAIII exhibits better convergence and diversity than those of the compared algorithms and yields an optimized VM placement scheme with shorter makespan, less user rental costs, and lower energy consumption.