Energy-Efficient Virtual Machine Dynamic Consolidation For Heterogeneous Cloud Platform

In recent years, cloud computing attracts extensive attention from academics and industries, and with the rapid development of cloud computing, large-scale data centers contain millions of computational nodes are built all around the world. Operations of those cloud data centers need a mass of energy consumption, which makes Cloud Service Providers bear high operating costs, produces large quantities of carbon dioxide and causes environmental pollution.The dynamic virtual machine consolidation technology is one of the effective methods to reduce the energy consumption of cloud data center. For this technology, multiple virtual machines are deployed on a single physical machine, and for the dynamic change of workloads on virtual machines, by using the on-line migration technology, it redeploys the virtual machines in order to improve the resource utilization of physical machines, reduce the number of active physical machines and improve the energy efficiency of data centers. Because that most of the modern data centers are composed of various models of physical machines, and they need to deal with various kinds of workloads, the heterogeneity of data centers and the multidimensional nature of the virtual machine resource required should be taken into account when designing dynamic consolidation algorithm of virtual machine. To solve above problems, the following research was launched in this paper:In the first part, the concept and basic framework of data center are introduced. Besides, the source of energy consumption in data center and basic pattern of the green cloud computing are analyzed. For the common virtualization technology used in cloud data center, several kinds of system virtualization methods are focused on as well as the on-line migration technology and process for different virtual machines.After that, we modeled the system of cloud data centers and energy consumption, and then built a system model of distributed cloud data centers to introduce the basic compositions of distributed cloud data centers, and analyzed and explained functions and treatment schemes of every module. We also modeled energy consumption of physical machines of cloud data centers to analyze energy consumption during transport process of virtual machines, and then got the best transport time.Finally the problem of dynamic integration of virtual machine is analyzed in cloud data center. Aiming at the isomerism of physical machine, an energy efficiency model is proposed to quantify the energy efficiency of each physical machine. For multidimensional resources of virtual machine, a resource utilization model is proposed to quantify resource utilization rate of the physical machine which has multidimensional resources. On the basis of problem analysis and each model which has been proposed, an integration algorithm of distributed dynamic virtual machine is designed. The algorithm consists of three parts:high/low load detection, choice of virtual machine and virtual machine placement. Compared with the existing virtual machine dynamic integration algorithm, the algorithm proposed in this paper has the...