Research On Cloud Environment Virtual Machine Allocation Optimization Strategy

With the rapid development of the Internet, the traditional computing and storage mode cannot meet the needs of individuals and enterprises. So the concept of “cloud computing” was proposed. Cloud computing subverts the traditional resources management, cloud computing allows users to access a virtual pool of resources through the network and applies for resources according to their needs. Cloud computing has been widely studied and applied for its distributed, dynamic, demand independent,low-cost and so on. How to efficient and reasonable schedule the resources of cloud data centers, and how to improve resource utilization is the important issue of the current research.This thesis mainly focuses on the scheduling algorithm of cloud data center resource. Based on summarizing the previous research, using economic theory and more efficient resource scheduling algorithm is proposed to greatly improve resource utilization and user utility. The main research work includes the following:Firstly, the thesis proposes the utility optimization resource scheduling algorithm based on the production function. Taking into account user satisfaction, the resource scheduling algorithm is to maximize the benefits of the cloud service providers. There are two stages in this scheduling solutions. In the first stage, cloud server agents should reasonably plan all resources in cloud data center to optimized allocation resources. In the second stage, cloud server agents allocate resources to cloud users according to users' requests. After this two stages, the algorithm improves the cloud data center resource utilization and revenue of cloud service provider.Secondly, this thesis also proposes total utility optimization resource scheduling algorithm based on “social welfare”. The resource scheduling algorithm is to maximize total utility value of all users. Firstly, the algorithm statistics and counts the ratio of the users' request resources and the total resources of data center, the ratio of the physical machine resources and the total resources of data center. Classifies users and physical machines according to the maximum value of the ratios; then, the utility function is built to calculate the utility value of the user to obtain the specific virtual machine, virtual machines are allocated to users according to the highest utility value. After the algorithm, the virtual machines utilization is improved and the user utility is optimized.Finally, the proposed algorithms are verified on the OpenStack cloud platform. The experimental results show that the proposed algorithms have a good performance in improving the resource utilization, optimizing user satisfaction and improving the revenue of cloud service provider.