Research On Dynamic Migration Strategy Of Virtual Machine Based On Load Prediction In Cloud Computing Environment

Cloud computing provides its own resources as a service to users,and users can rent and publish them on the Internet in an on-demand manner.Currently,Infrastructure as a Service(IaaS)is one of the widely used cloud computing service models.It provides cloud computing resources to users in the form of virtual machines(VMs),encapsulates their resources as cloud computing services to consumers on demand.Fee.Virtualization technologies such as Xen and VMware allow infrastructure resources to be shared in a useful way.Virtual machines can also dynamically allocate cloud computing resources based on different needs,provide opportunities for efficient use of cloud computing resources,and optimize application performance and energy consumption.Dynamic load balancing in cloud computing can be implemented by migrating virtual machines from heavy or lightly loaded hosts to other hosts to save energy and mitigate performance-related Service Level Agreement(SLA)conflicts.The two main issues in cloud load balancing dynamic load balancing are how to predict heavy load hosts and light load hosts.The second is how to develop a virtual machine migration strategy.This paper is devoted to the study of virtual machine dynamic migration strategies based on load forecasting under cloud computing environment.By regulating multiple resource utilization parameters under cloud computing environment,the host can be overloaded or lightly loaded,and then the virtual machine migration strategy can be used to implement cloud computing.Data center load balancing.This paper first optimizes an existing load forecasting algorithm.The second-order cone programming formula is mainly used to optimize the robustness of the load prediction algorithm,in which the training mode is represented by an ellipsoid rather than a simplified convex hull.The first is to derive a linear formula,and then the kernel-based method is to construct a decision model for nonlinear classification.Finally,experiments on the data set show that the optimized algorithm has improved performance.Secondly,this paper compares the optimized algorithm with other load forecasting algorithms,uses the most accurate load forecasting algorithm in the test model,and uses the correlation coefficient method to consider the relationship between the virtual machine and the virtual machine in the host,and the relationship between the virtual machine and the host.VMs that need to be migrated to implement virtual machine dynamic migration strategies.