Research On Energy-Efficient Scheduling Model And Algorithm In Cloud Computing Environment

With the development of cloud computing,data centers have become the footstone of global economic development.Therefore,data centers have developed rapidly in terms of scale and growth rate.However,a large number of data centers have caused huge consumption of power energy.At the same time,it has increased the economic cost of cloud computing,and led to the soaring emissions of carbon dioxide,which has an unimaginable impact on the global climate.Therefore,energy consumption has become an important issue in cloud computing research.In order to achieve the effective management of energy consumption and the rational allocation of resources in cloud data center,we study the energy-saving scheduling model and algorithm in cloud computing environment,which is of great practical significance.We make a survey on energy consumption aspects,including energy saving technology in data center,energy consumption model and task scheduling algorithm,which addresses huge energy consumption and low energy efficiency in data center.Then,we analyze the existing problems and difficulties of energy consumption model and task scheduling algorithm,and carry out the following work:(1)To achieve energy saving and emission reduction,the accurate measurement of energy consumption is needed.Energy consumption model is an important tool to measure the energy consumption.The heterogeneity of modern data centers often leads to huge differences in energy consumption,which puts forward high requirements for the accuracy of energy consumption models.Therefore,we combine traditional energy consumption monitoring technology with virtualization technology,and propose a non-linear energy consumption model based on virtual machine's CPU frequency and core in heterogeneous cloud environment(HFNM).Considering the heterogeneity of the data center,it provides a new solution to the energy consumption monitoring problem of cloud computing data center.The experimental results show that HFNM can provide accurate energy consumption estimation for virtual machines under computing intensive workloads in that the model reduces the errors caused by heterogeneity of data centers.(2)The quality of scheduling algorithm is related to the performance and stability of data center,which is a significant issue for cloud computing.Due to the diversity of user requirements,heterogeneity and complexity of data centers,there are many problems in task scheduling,such as low quality of user service and high energy consumption.On the basis of the non-linear energy consumption model HFNM,we further propose an energy-aware multi-objective task scheduling algorithms(EMTS)for computing intensive loads with task deadlines as constraints to reduce total energy consumption.The simulation results by CloudSim show that the proposed algorithm works well.EMTS is superior to the traditional polling algorithm in terms of system energy consumption,total completion time and task timeout rate.