Research And Implementation Of The Deployment Method Of Virtual Machine Based On Cloudstack

Cloud service providers use virtualization technology provide the compute, network, storage and other resources as a service to users trough the network. These resources constitute the cloud data center resources pool. In IaaS service model, a single physical host in the resource pool is virtualized into several or even hundreds of virtual machine. The background resource management node use the virtual machine to manage the virtual resources of the resource pool. Users using virtual machine to complete a series of tasks is actually a process of allocating an scheduling resource. Rational resource management can impove resource utilization and reduce energy consumption. With the development of cloud computing technology, users demand for resources is increasing. The number of servers in the resource pool is growing. Unreasonable virtual madhine deployment will generate a lot of resource fragments and make the problem of waste of resources more serious. So, how to avoid wasting resources and how to reduce resource fragments is an urgent problem.To solve the above problems, the paper takes the aim at reducing the fragments of resources, to propose a new virtual machine deployment method.1. On the basis of summaring the research achievements of home and abroad scholars, the paper designed the new virtual machine deployment method. The method is divided into three modules, which are named prediction module, planning module and adjustment module. Prediction module considers the users’behaviors, it predicts the incremental of different specifications of the virtual machine in the next period time by the historical data of the users applicating virtual machines. The planning module dynamicly plans the available resources of resource pool bases on the data which is obtained from the prediction module. The program module sets the resource allocation, so that when the user applies a virtual machine, the background resource management server can directly deploy the virtual machine on a specify physical host. When the setted resource allocation rule no longer fit the user to use, triggering the adjustment module, predicting the increments of the virtual machine again.2. The paper achieve the virtual machine deployment method on the cloudstack cloud computing platform provided by some telecoms Operator. Computing the fragmentation rate of the resources on the physical host and compares with the fragmentation rate of three deployment methods, one of which is used by the telecoms Operator. Through the experiments, the result verifies the proposed deployment method of virtual machine can reduce the resource fragments effectively.