Research And Implementation Of Virtual Machine Management Mechanism On Campus Cloud Platform

With the extensive application of cloud computing technology, which is to solve some problems of low utilization rate, high cost of maintenance, information isolated island in the infrastructure construction of the traditional data center provides a better solution—ampus Cloud.However in the practical application environment, due to server load was too high, which leads to a large number of dynamic virtual machine migration in a certain period of time, which will not only waste of network bandwidth resources, and causes the server instability caused by the interruption of business. Analysis of the reasons that Campus Cloud is different from the traditional cloud platform, the traditional cloud platform bear the run of virtual machines of the key business systems, The virtual machines cycle is long, high stability. According to the application characteristics of Campus Cloud, it is divided into two application types of scientific research and teaching. The running state of virtual machines of scientific research type often by the researchers, the virtual machine needs in a certain period of time is stable, has the characteristics of non-periodic. The virtual machine of teaching type operation is usually decided by the teaching schedule, with periodic characteristics Therefore, according to the need of virtual machine scheduling mechanism design features and requirements of the reasonable.This article from the perspective of the virtual machine running time analysis the common characteristics of scientific research and teaching, with the goad of load balance, distribution load to each computing node, to avoid a large amount of virtual machine migration by reason of uneven load of CPU and memory each node. The specific research work and the main contributions are as follows:(1) Put forward the theoretical model of scientific research and teaching application of virtual machine scheduling mechanism in Campus Cloud platformFirst, abstract the application requirement of teaching and scientific research, put forward resource model, time model, demand model, definite concept of load rate and load set etc.(2) Put forward virtual machine scheduling algorithm for scientific research and teaching in Campus Cloud platform.This paper considering the scientific research and teaching application virtual machine load characteristics, in order to solve the traditional virtual machine scheduling mechanism caused by the cluster load problem, "first to calculate the physical machine load prediction, and balanced deployment principle and" to load balance as the goal, low physical machine high load dynamic scheduling strategy is proposed.(3) The design of the Campus Cloud platform system architecture and system implementationThe Campus Cloud platform system logical architecture is designed according to the proposed virtual machine scheduling mechanism. The architecture includes platform interface, resource management and service, operation and maintenance monitoring, resource interface, hardware and software infrastructure. According to this logic architecture, build a high availability, scalability, multi node OpenStack cloud platform, and ultimately through the Dasein Cloud open source interface to achieve the DeCloud-chasis campus cloud platform virtual machine management system.Experiments show that the proposed virtual machine scheduling mechanism to ensure the teaching and scientific research and application needs, and make full use of cloud host resources, to achieve load balance between the cloud host.