Research On Memory Pre-copy Algorithm And Energy Saving Strategy Based On Live Migration Of Virtual Machine

With the rapid development of computer technology,virtualization has become more useful.Because the live migration of virtual machines is one of the key technologies of virtualization,the research on the live migration of virtual machines is of great significance.On the one hand,in the live migration of virtual machines,memory migration has become a hot topic in the study of live migration of virtual machines due to the large amount of data transmission and the complex migration mechanism.At present,the traditional memory pre-copy(MPC)method that is mainly used has the problem of retransmission of high-dirty pages,resulting in a serious drop in migration efficiency,and it is urgent to propose a new MPC algorithm to solve this problem.On the other hand,because the virtual machine will generate a lot of high energy consumption during the start-up and shut-down phases,it is also urgent to propose an energy-saving strategy to reduce the high energy consumption and reduce the energy waste phenomenon.This paper focused on the premise of how to reduce the migration time,reduce the amount of data transmission and reduce the energy consumption,and conducted a memory pre-copy algorithm and energy-saving strategy based on virtual machine live migration.This paper analyzed the memory migration in virtual machine live migration and designed a memory pre-copy algorithm based on Dirty Page Rate Prediction(MPRA_DPRP).The MPRA_DPRP algorithm predicts the dirty page rate of the virtual machine memory page based on the dirty page rate prediction theory,and identifies high dirty pages from the dirty pages generated during the migration process.During the iterations,small changes in memory pages are forwarded to prevent high dirty page retransmissions from affecting the overall performance of the migration.The MPRA_DPRP algorithm implemented the process of virtual machine memory migration in Xen,and according to the entire migration time,downtime and data transmission value,the experimental results were compared in three aspects of different CPU load,memory load and network load test.The experiments in this paper showed that the MPRA_DPRP algorithm can reduce the amount of data transfer by about 100 MB compared to the previous MPC algorithm,and the total migration time is reduced by nearly 19%,which effectively improves the efficiency of memory migration in virtual machine live migration.During live migration of virtual machines,this paper proposed a Live Migration Algorithm with Low SLA Defaul Rate(LMA_LSLADR)to reduce energy consumption and maintain a low SLA default rate.Specifically,in order to achieve less energy consumption and maintain a small SLA violation rate,this article adopted a live migration method based on two thresholds in the triggering phase of virtual machine live migration: during virtual machine selection and allocation period.The strategy of introducing the minimum migration cost starts the migration of the virtual machine;in the selection stage of the target host,the target node is selected in a manner based on a probability selection mechanism.From the three aspects of energy consumption,number of virtual machine migrations,and SLA default rate,compared with several traditional live migration methods,the CloudSim experiment showed that the LMA_LSLADR algorithm can make system resources more efficient,less consumption and save energy.