Research On Virtual Machine Deployment Algorithm Based On Energy Saving And Prediction

Cloud computing service providers rely on data centers with a large number of physical machines to meet the needs of users.However,maintaining such an infrastructure generates a lot of energy consumption.Virtual machine deployment(including initial deployment and re-deployment)is an effective way to reduce energy consumption in the data center.In recent years,the issue of virtual machine deployment has received a lot of attention and research.At present,relevant research on virtual machine deployment algorithms has achieved certain results,but there are still some deficiencies.In order to solve these problems,this thesis studies the deployment algorithm of virtual machines,and proposes a virtual machine deployment algorithm based on energy saving and prediction.The algorithm includes two parts: the initial deployment algorithm and the redeployment algorithm.The research work of this thesis are as follows:1)Based on virtual machine pre-copy migration technology,a new method for calculating the migration cost of virtual machines is proposed.This method considers three factors: migration energy consumption,downtime and migration time,which can more accurately measure the migration costs of virtual machines.At the same time,the proposed calculation method is more accurate in measuring the migration costs of virtual machines.2)In-depth study of virtual machine deployment and virtual machine resource prediction research status is conducted.Based on the migration cost and the prediction of virtual machine resources,a re-deployment algorithm of virtual machine is proposed: MCA.The MCA algorithm first uses the Wiener filter predictor to predict the virtual machine resource.Then,based on the predicted results,the redeployment of virtual machines is performed considering the impact of the number of active physical machines and migration costs(migration time,downtime,and migration energy consumption).Compared with the existing virtual machine redeployment algorithm,this algorithm reduces the energy consumption of cloud data centers and reduces the migration cost.3)Proposing a virtual machine initial allocation algorithm —DRMGA(Different Resource with Minimum Gap Algorithm).DRMGA considers both the task type and the use of different resources of the physical machine.In the initial deployment of a virtual machine,tasks are categorized according to the task's demand for different resources.And then,different deployment strategies is used according to the type of task and the use of different resources of the physical machine.Compared with the existing initial allocation algorithm of virtual machines,this algorithm can reduce the number of physical machines in the cluster and reduce the energy consumption in the cloud data center to a certain extent.4)The initial deployment algorithm and re-deployment algorithm of the virtual machine proposed in this paper are verified by experiments and compared.