Virtual Machine Migration Strategy Based on the Load Evaluation and Prediction under the Cloud Environment

Concerning the centralized communication overloading of the virtual machine (VM) in the cloud computing center, this paper puts forward an adjustment strategy, that is, to build a model through the load evaluation of three threshold values in consideration of the load balancing and to achieve the objective of dynamic migration of the VM. First, load evaluation algorithm for four statuses is designed by increasing the division grade of the physical machine (PM). The exponential smoothing model based on time series trend prediction is introduced. Through the prediction of the load in the next moment, the overloaded PM is certified. The VMs to be migrated and the optimal objective target hosts are found out through the calculation of the utilization rate and the utilization trend of various VMs, CPUs, internal storage and bandwidth of target hosts. Based on that, the model is established and the VMs are migrated.