Research On Virtual Machine Placement Strategy In Cloud Computing

Cloud computing is based on virtualization technology and use pay-as-you-go as the business model, and is a novel computing paradigm with features like elastic scalability, dynamic allocation, resource sharing and so on. In the cloud computing model, hardware, platforms, software and other IT resources are provided as a service to users. According to the level of services provided, cloud computing can be divided into three levels: IaaS, PaaS, SaaS. In the IaaS layer, with the system virtualization technology, the servers in data centers are virtualized, and can achieve improvement of resource utilization, reduce energy consumption and simplify deployment by using virtual machine as the granularity for resource management. In the virtualized resource management, virtual machine’s deployment is an important step, and the decision of the host’s selection in it are directly related to whether the resources of the servers and network equipments can be effectively used, and inadequate use of resources will produce additional energy. Besides, with the development of cloud computing, in order to meet the user’s growing resource demands for cloud services, capacity and number of the servers and network equipments of the data center are increased which will also generate a lot of energy. A lot of energy can result in increasing costs and environmental pollution problems. Therefore, how to reduce the energy consumption of data centers has become an urgent problem.This thesis aims to address the above problem. With the goal of reducing the energy consumption of data centers and from the perspective of optimizing the placement of virtual machines, the virtual machine placement policy of reducimg the energy consumption is proposed based on the in-depth analysis and study of the existing virtual machine placement strategies. The main contents are as follows:1. A new virtual machine placement policy which based on performance vectors is proposed by researching the methods of reducing servers’ energy consumption and modeling the virtual machine placement problem of considering energy consumption of servers. The policy takes into account the multidimensional resource, heterogeneous consumption of servers, applications’ individual needs and other factors, and is able to select a server with the minimum amount increase of energy and more consistent resource status with the application’s resource requirement for the application’s corresponding virtual machine. Based on the proposed policy, the corresponding algorithm has been put forward and validated by doing simulation experiments.2. Based on the above research achievements, a new virtual machine placement policy which considering network equipments’ energy consumption is proposed by furtherly considering the energy consumption of the network devices in the data center, and modeling the virtual machine placement problem considering network equipments’ energy. The policy considers the traffic demands of application’s corresponding virtual machines and tends to minimize network equipments to complete the routing of traffics and takes into account the energy consumption of servers to achieve the global minimum of the data center servers and network equipments. Based on the proposed policy, the corresponding algorithm has been put forward and validated by doing simulation experiments.