Design Of Adaptive Compression Optimization Model For Live Migration Of Virtual Machines Based On Bandwidth

Virtualization technology,as one of the core technology of cloud computing,has a wide range of applications in the field of information technology.virtualization technology is a logical abstraction of physica l resources,while is easy to achieve a integration of network servers and provide a powerful computing power.Whether the cloud service provider or virtual open source community are working for further improvement of virtualization technology,the virtual machine live migration technology is a major highlight of virtualization technology,providing a effective support for server virtualization.The live migration technology can move the virtual machine to another host physical machine without interrupting the client application.It plays an important role in load balancing,disaster recovery,resource utilization,energy consumption and other aspects of data center.The live migration technology is one of the key technologies indispensable to data centers or cloud platforms.With the continuous expansion of the Internet business,the current virtual machine live migration technology is not well adapted to the complex environment,as it still exists a long migration time,large amount of data migration and so on.So this topic of my thesis is based on the virtual machine live migration problem in-depth study.The focus of this research is to understand the live migration mechanism of the mainstream virtual machine based on the KVM(Kernel-based Virtual Machine),and to study the live migration process of the virtual machine in depth.Virtual machine live migration process is mainly on the memory data migration.For memory-intensive applications in the use of virtual machine,pre-copy of the livemigration mechanismwill increase the frequency of dirty pages;memory dirty pages will be frequently copied.This may not only increased the number of iterations,but also greatly increased the network burden.In order to reduce the migration time and the amount of migration data,this paper proposes an adaptive compression optimization model of ACMB(Adaptive Compression Model based on Bandwidth)based on bandwidth.This model introduces two different compression strategies in the face of high bandwidth and low bandwidth.One of them focuses on compression ratio,named Zstandard,and the other is the fast compression algorithm,which focuses on the compression speed,named LZ4.It can be based on the different bandwidth of the dynamic selection of the appropriate compression algorithm.The optimization model first determines the network transition bandwidth threshold of the two compression algorithms,then obtains the real-time bandwidth of the network and passes it to the compression algorithm pool,then selects the appropriate compression algorithm according to the critical point,and then optimizes the selection of the compression algorithm by finding the optimal acceleration level.Finally,the performance test of live migration is carried out on the pre-optimized and optimized model,and the results are compared with the three indexes of total live migration time,total live migration data and downtime.The results show that the ACMB model proposed in this study has a significant improvement in the thermal migratio n performance of the virtual machine.