The Research On Dynamic Forecast Management Algorithm For Virtual Machine In Cloud Computing Environment

Cloud computing is based on the modern of the increasment of the related services,use and delivery of the Internet,usually using the Internet to provide the dynamic and easy extension and virtualized resources.The user can get efficient and reliable and economical computing resources by widespread application without additional costs,such as the purchase and maintenance of facilities,daily management,etc.This application mode of innovation benefit from the wide application of virtualization technology.At the time,with the widespread application,the virtual resources grow rapidly how to efficient management of virtual resources,improve efficiency become the important fields of research.This paper study from the virtual machine migration and virtual machine deployment in cloud,in order to improve the utilization rate of resources.At first,considering of the three elements of CPU,memory,disk resources,the virtual machine deployment problem is described as the packing problem,we analysis the statistical consumption of CPU,memory,disk resources in virtual machine with the increasing time,finding the dynamic change of consumption values,in the case,by designing the time series model predict the next cycle of resource consumption value of the CPU,memory,disk resources depend on the history consumption values.then we put forward a new virtual machine deployment algorithm VMPA(virtual machine placement algorithm).the application value of the physical server CPU,memory,disk resource can be allocated the threshold,the virtual machine value of CPU,memory,disk resources in servers do not exceed the threshold value of the prediction server,it can deploy.To make better use of resources and ensure that the system satisfies the requirement ofSLA(service level agreement),migrating some virtual machines from some nodes which under the heavy load,considering for CPU,memory and using AHP(analytic hierarchy process),calculating the value of the load on the server,and setting up the load value meeting the SLA,and using the time series to predict the next cycle of resource consumption value of the virtual machine CPU,memory as the basis of migration and placement of the migration.Then the paper put forward a virtual machine migration algorithm VMMA(virtual machine migration algorithm),in order to meet the system requirements of SLA(service level agreement),migrating the appropriate virtual machine from the hot spot which the load is heavy to the appropriate physical server.and choosing a appropriate server to deploy the virtual machine.When the virtual machine would be migrated to the physical server,the disk resources would set the default application value and the value of CPU and memory is the origin value in last server.Experiments prove that two algorithm can effectively improve the utilization of resources.