Research On Virtual Machine Migrationoptimization From The Perspective Of Granular Computing

As a new commercial computing model of on-demand service,more and more professionals will welcome cloud computing.With the advancement and popularization of cloud computing technology,data center energy consumption has become more prominent and severe.The main reason is that the resource utilization of the cloud data center is insufficient,and the host load is unbalanced.Therefore,the improvement of cloud datacenter resources is insufficient,and the problem of imbalance is one of the main ways to save energy.In cloud computing,virtual machine migration technology can allocate and reorganize virtual machines and host resources in the data center,thereby improving the utilization of host resources,which is beneficial to host load balancing and achieves the purpose of reducing data center energy consumption.This subject conducts a relatively comprehensive survey of virtual machine migration technology,expounds on the current research status of virtual machine migration technology,and focuses on analyzing the factors considered in the migration process of virtual machines,and then optimizes virtual machine migration.The study found that a host with a high load greatly impacts energy consumption and performance.Therefore,researching the host according to the load situation can help balance the host load.This idea is consistent with the granular computing theory,and the three decisions are used as the granular computing field.To solve the problem,using the basic idea of three-way decision to introduce the virtual machine migration strategy,we propose a series of optimization schemes to solve the energy consumption problem.(1)To achieve data center load balancing and reduce energy consumption,this subject proposes a virtual machine migration strategy based on three decisions.According to the three-way decision model,the hosts are divided into high-load hosts,normal-load hosts,and low-load hosts.Considering that the load and resource utilization characteristics of different types of hosts are different,different factors are considered in the virtual machine migration strategy for different hosts.The high-load host considers the multi-dimensional resource balance of the host after the virtual machine is pre-migrated.Considering the two factors of the decrease in machine load,the low-load host considers the multi-dimensional resource balance factor of the virtual machine after pre-placement of the virtual machine.Through the migration of high-load hosts and the migration of low-load hosts,the data center will eventually be owned the load of the host is converted to the normal load range,realizing the load balance of the host and reducing the energy consumption of the data center.(2)Aiming at the problem of network prerequisites between virtual machines,that is,because different virtual machines may perform different tasks of the same job,and there is data communication between tasks,network communication will occur between virtual machines,so the virtual machine migration process May need to consume additional network communication overhead.In response to this situation,this topic proposes a virtual machine migration strategy based on network correlation,which divides the host into heavy overload,medium overload,light overload,medium load,ultra-low load,medium and low load,light and low load in a more granular manner.There are seven host types.Simultaneously,virtual machine migration is divided into immediate migration,delayed migration,and no migration,and targeted virtual machine migration decisions are taken for different types of hosts.Finally,the experiment was verified through the Cloud Sim cloud simulation platform.The results show that compared with other virtual machine migration strategies,the strategy proposed in this topic effectively maintains the load balance of the data center host and reduces the energy consumption of the data center.