Research On Scheduling Algorithm Of Virtual Machine With Low Energy Consumption In Cloud Data Center

The global COVID-19 has accelerated the development of cloud computing.Business requirements related to cloud computing are increasing.This undoubtedly brings unprecedented pressure to the cloud data center,and at the same time,the excessive energy consumption of the data center also hinders the development of cloud computing.The development of virtualization technology provides a better way to solve the problem of data center energy consumption.Virtual machine migration and integration can not only balance the resource utilization of the data center,improve the quality of user experience,but also have important significance in reducing the energy consumption of the data center.At present,there are many virtual machine scheduling strategies based on energy saving,and good results have been achieved.However,most virtual machine scheduling strategies only use real-time load conditions as the basis for virtual machine placement,ignoring the impact of energy consumption on the system after virtual machine migration.This paper proposes a low-energy scheduling strategy for virtual machines based on load prediction,aiming to construct a strategy for virtual machine scheduling with high prediction accuracy,balanced utilization of physical machine resources,and fewer migrations.The research content of this paper mainly includes the following aspect:(1)A composite prediction model is proposed to predict the utilization of physical machine resources in a period of time in the future.Use the compound prediction model of ARIMA method and cubic exponential smoothing method to predict the time-varying physical machine load sequence,use the ARIMA model to mine the linear relationship,eliminate the random fluctuation of the physical machine load sequence,and use the cubic exponential smoothing method to mine the nonlinear relationship.Smooth the physical machine load sequence.The weights of the two models in the composite model dynamically update the weight coefficients of the composite model according to the prediction accuracy of the ARIMA model and the cubic exponential smoothing model.The experimental results show that the model proposed in this paper can predict the load sequence more completely.(2)An improved NSGA-Ⅱ multi-objective optimization algorithm is proposed.When selecting individuals using the elite selection strategy,the normal distribution function is used to control the number of optimal solutions in each layer to ensure uniformly distributed solutions.At the same time,the matching function is introduced to balance the utilization of various resources of the physical machine when calculating the individual fitness.The experimental results show that compared with the improved particle swarm algorithm and NSGA-Ⅱ algorithm,it can balance the utilization of various resources of the physical machine and reduce energy consumption.(3)A virtual machine low-energy scheduling strategy combining load prediction and improved NSGA-Ⅱ algorithm is proposed.In the virtual machine scheduling process,predictive sequences and multi-objective optimization algorithms are used to select the target physical machine,reducing the number of migrations and reducing the energy consumption of the data center.The simulation experiment proves that the compound prediction model proposed in this paper can predict the linear and non-linear relationship of the physical machine load sequence better than the single prediction model.Secondly,compared with the traditional virtual scheduling algorithm and the improved intelligent heuristic algorithm,the method proposed in this paper can reduce the number of virtual machine migrations and the number of physical machines running,can better balance the resource utilization of physical machines,and effectively reduce data energy consumption.