Power Aware Dynamic Migration Algorithm Of Virtual Machine For Cloud Data Center Network

With the rapid development of mobile Internet,cloud computing technology is widely used because of its on-demand payment and flexible configuration,which can effectively reduce the cost of Internet enterprises.However,with the rapid development of the cloud computing industry,the scale of the data center continues to expand,and its demand for energy is increasing,resulting in extremely high energy consumption.The extremely high energy consumption is generated primarily for two reasons: low resource utilization in the data center and narrow server dynamic power range.The low resource utilization of the data center is due to the lack of effective resource management mechanism to deal with the dynamic changes of data center resources.The server's narrow dynamic power range is reflected in its hardware characteristics: the server's static power consumption is about 70% of the peak power.By migrating and consolidating virtual machine instances in multiple physical hosts,the virtual machine dynamic migration technology can effectively reduce the number of active physical hosts,and then shut down or put idle hosts into sleep mode,which can improve data center resource utilization and reduce the energy consumption of the data center.However,excessive migration and integration of virtual machines can cause the physical host to be overloaded,thus violating the user's Quality of Service(Qo S)requirements.As cloud computing users have different requirements for Qo S,it will also affect the migration strategy.Therefore,under the premise of ensuring the Qo S requirements of cloud computing users,how to design the virtual machine dynamic migration algorithm with full consideration of the Qo S requirements of different users has become the main research in this paper.The main work of this paper is as follows:1.Aiming at the problem that the energy consumption models in existing researches are too ideal,considering the non-linear power consumption characteristics of the central processing unit(CPU),a more realistic virtual machine dynamic migration energy consumption model is established.By considering the virtual machine migration energy consumption,the overall energy consumption model of the data center is improved.On this basis,this paper establishes a mathematical model of virtual machine dynamic migration problem with the goal of minimizing the overall energy consumption of the data center,and proves the NP-hard characteristics of the problem.Further,the time complexity of solving the problem accurately under different scales is given by simulation.2.An adaptive virtual machine dynamic migration algorithm based on user Qo S requirements and system load federation sensing is proposed.Firstly,considering the difference of Qo S requirements of virtual machines and system load,a method for determining the overload threshold of physical hosts is proposed.Then,a Qo S requirements aware virtual machine dynamic migration algorithm is proposed.By migrating virtual machines with similar Qo S requirements to the same physical machine and adaptively adjusting the migration threshold as the system load changes,the purpose of improving the utilization of physical host resources and reducing the number of active hosts is achieved.Furthermore,based on the energy consumption model established in this paper,the dynamic threshold of underloaded physical machine is designed,and an energy cost aware best fit algorithm is proposed,which is to migrate the virtual machine instances from low-load hosts to high-load hosts so as to reduce the number of active hosts and minimize the energy consumption.3.Based on the Cloud Sim cloud computing simulation platform and the extension of the platform,the virtual machine dynamic migration algorithm proposed in this paper is simulated under different request arrival rates and different data center scales,and the simulation results are compared with several existing methods.The simulation data shows that the proposed algorithm is superior to the existing algorithms in terms of system energy consumption,request acceptance rate and number of active hosts.In addition,by finding the optimal solution of the problem on a small scale,the difference between the proposed algorithm and the optimal solution is given,and the superiority of the proposed algorithm is illustrated from the perspective of time complexity.