Research On Dynamic Management Mechanism Of Virtual Machine In Cloud Computing Data Center

Cloud computing is a model for enabling convenient,on-demand network access to a shared pool of configurable computing resources(e.g.,networks,servers,storage,applications and services).As the core of the cloud computing,the performance of data center has a significant influence on the upper services.However,there are many problems in cloud data center,such as high power consumption,low resource utilization and unbalanced resource usage.To solve these problems and improve the overall performance of the data center,dynamic management of virtual machine in data center is studied in this paper.The virtual machine dynamic management includes three processes:host migration timing decision,virtual machine selection and virtual machine placement.Therefore,the dynamic management mechanism is studied from host migration trigger strategy,virtual machines selection strategy,and virtual machines dynamic placement strategy in this paper.The core of the host triggering migration is host state detection.In this paper.we propose a two-level state detection strategy based on sliding window and a state prediction model based on Holt exponential smoothing method,aiming at eliminating the influence of volatility and trend on host state detection and improving the effectiveness of host state detection at the same time.In order to select optimal virtual machines to migrate.multiple evaluation indicators including the virtual machine migration gain,the virtual machine migration cost and the virtual machine resource competition are designed in this paper,taking the factors such as the size of the host resource occupied by virtual machine,the virtual machine migration time,the virtual machine performance loss and the virtual machine resource utilization rate into account;And then we use a multi-objective decision analysis method to balance the conflict of multiple indicators.Aiming at the shortcoming of optimizing only some of QoS.power consumption and resource utilization in the previous studies,all of these goals are considered in this paper when placing the virtual machine,and a virtual machine dynamic placement algorithm named DPKS is proposed,which is more suitable for the large-scale growth of data center because of a lower time complexity.The virtual machine dynamic management system of cloud data center(VDMCloud)is designed and developed based on OpenStack,then deployed and tested in Sugon cluster.The test results show that the models.algorithms and strategies proposed in this paper can significantly reduce the power consumption of the data center,improve the resource utilization rate and make the resources relatively balanced under the premise of guaranteeing the QoS.which improve the data center performance and make a contribution to the development of data center management.