Research On Strategy Of Virtual Machine Resource Allocation And Placement In Cloud Computing Environment

Cloud computing has changed the traditional computing resources delivery model, which is the research hotspot presently. Due to the characteristics of providing resources on demand, pay-as-you-go manner and dynamic scalability in the cloud computing, there are many challenges in the cloud computing research, especially in the virtual machine resource allocation, deployment, migration and the integration of physical server, etc. Virtual machine resource allocation and deployment are the two core issues in the cloud computing research. In the traditional research of virtual machines resource allocation and deployment based on a unilateral consideration, ignored some parameters which concerned by users and cloud service providers. Therefore, the allocation of virtual machine resources in the cloud computing market and the deployment of virtual machine in the data center has great academic and practical value.For the issue of virtual machine resource allocation, the cloud service provider in the cloud computing market is the object of research in this thesis, we introduce the evolutionary game theory in economics, modeling and analysis of virtual machine resource allocation in cloud computing. In the entire cloud computing market, the game strategy are price and quality of service of virtual machine resources published by cloud service provider. Evolutionary game based on the price and quality of service, in the game process, cloud computing service providers adjust the game strategy dynamically in order to improve the utility. Finally, we proved that the entire virtual machine resource allocation process to achieve evolutionary stable from both theoretically and experimentally.For the issue of virtual machine resource deployment, the data center of cloud service provider is the object of research in this thesis, the Particle Swarm Optimization is theoretical guidance, modeling and analysis of virtual machine resource deployment in data center. Quantify the attribute of CPU, memory, storage of each physical machine and virtual machine. According to the demands of users and the current load status of physical server, we adopt multi-objective particle swarm optimization algorithm to optimize the deployment of virtual machines based on the physical server resource utilization and the number of virtual machine migration of these two objectives. Simulation results show that the algorithm can reach the minimal number of virtual machine migration and improve resource utilization effectively.