Research On Energy Comsuption Optimization Of Cloud Data Center Under SDN Architecture

With the rapid development of information technology and Internet of Things technology,the ICT industry is maturing.Ultra-large-scale data centers must be established to store and process massive amounts of data.Data centers usually host tens of thousands of physical hosts,and these hosts are interconnected to form a huge network and exchange information with the outside all year round.Therefore,the rapid expansion of scale,load pressure,storage pressure and energy consumption problems are increasing more and more seriously.According to statistics,the data center will become the world’s largest energy consumer in 2025.In addition,despite of the environment of global energy shortage,the carbon emissions of data centers are reach up to 2%.The redundancy of energy consumption in data centers has become one of the most important issues that restrict the development of operations,hinder the reduction of operating costs,and respond to energy-saving and emission-reduction policies.As a result,how to reduce energy consumption and carbon emissions as much as possible on the basis of ensuring the normal operation of data centers has become a hot issue.This article mainly studies the energy efficiency optimization of the data center from three aspects.1.Research on energy efficiency optimization of data center servers.The rigidity of the current data center static network structure failed to fully perceive the data center network and equipment dynamics,and deploy and implement strategies inflexible and unthorough,this paper introduces the SDN architecture as the system architecture,and proposes an energy consumption optimization algorithm composed of an energy-driven virtual machine migration scheduling algorithm and a dynamic scheduling based on Bayesian method named as DSSBM algorithm,and then deployed it as SDN data center energy-saving strategy algorithm.This strategy makes use of the advantages of separation of network control and forwarding in the SDN architecture.It dynamically senses the network load and virtual machine load of data center in real time,dynamically schedules tasks,improves the utilization of virtual machines under working conditions,and makes as many idle virtual machines enters the sleep state as possible.The simulation results show that the performance of this strategy is better than other typical energy optimization algorithms.2.Research on energy efficiency optimization of data center network.The Fat-Tree topology architecture is introduced to the software-defined data center network,and a dynamic task scheduling strategy SGAFRS is proposed.The algorithm is based on genetic algorithm to optimize the energy consumption of switches,and deploying the data center network through task path selection algorithm.This strategy reduces the energy consumption of switches and links by improving the energy utilization of switches,reducing the number of transmission hops of tasks,thereby reducing the energy consumption of the data center network.The simulation results show that the energy saving of SGAFRS can achieve 24% in the best time period.3.Research on optimization of data center energy dispatch and energy supply structure.The power supply system and data center adopt the SDN architecture as the system architecture.An energy consumption prediction algorithm PsGMcR algorithm based on GM(1,1)energy prediction and Markov chain error correction is proposed,and an energy scheduling strategy EDS is proposed based on the PsGMcR algorithm.In addition,this strategy introduces solar energy to the power supply system,so as to achieve the purpose of energy structure optimization,energy saving and environmental protection.The simulation results show that the strategy is better than the benchmark algorithm and effectively reduces the carbon emissions of data center.The energy cost saving rate is about 12%.