The Resource Scheduling Approach For Cloud Energy Consumption Optimization

As an important embodiment of modern high-tech development,cloud computing has become a powerful engine for the development of innovative enterprises.With the innovation of the service model of pay-as-you-go,cloud computing has attracted wide attention from enterprises and users.However,the rapid expansion of cloud platform has caused a series of high energy consumption problems,which has become a major problem faced by cloud service providers.How to effectively reduce energy consumption of datacentre while ensuring cloud service quality has become a hot issue in cloud computing field.As one of the important means of energy consumption optimization,cloud resource scheduling technology aims to design efficient resource allocation strategies to improve resource utilization and reduce energy consumption.Therefore,this dissertation focuses on energy consumption optimization in cloud environment,in-depth analysis and comparison of key issues of current cloud resource scheduling technology,with the help of theoretical tools such as three-way decision and granular computing,from the aspects of virtual machine migration technology,cloud task scheduling strategy and cloud load prediction methods,a series of resource scheduling methods for cloud energy optimization are carried out.(1)In view of the virtual machine resource competition and inefficient utilization caused by the limitation of a single physical machine in the actual cloud platform,by analyzing the virtual machine load similarity and resource occupancy problems,this dissertation proposes an energy consumption-aware virtual machine migration strategy based on the three-way decision.Firstly,it design a trisecting strategy of multi-granularity for virtual machine migration and use the K-means method to cluster for each division region and then select the virtual machine sequence that need to be migrated.Secondly,it calculate the similarity between the virtual and physical machine resources and place the virtual machin.Finally,a simulation experiment will be conducted on the CloudSimPlus platform,and the results show that this method can effectively reduce energy consumption and fully utilize resources.(2)In view of the high energy consumption loss caused by the mismatch between cloud tasks and resources in the process of cloud task scheduling,this dissertation introduces the multi-granularity knowledge discovery method into cloud task scheduling and proposes a cloud task scheduling method based on greedy strategy under the perception of energy consumption.First,the tasks are divided into different granularity:CPU type,RAM type and mixed type;Then assign corresponding scheduling strategies for cloud tasks with different granularity.Finally,the experiment was carried out on CloudsimPlus platform.The experimental results show that the proposed method can greatly reduce the energy consumption of cloud datacentre,and is a very practical task scheduling algorithm.(3)In view of the current situation that the cloud load demand prediction algorithm is difficult to meet user resource demands which is dynamic changing under the quasi-periodic effect.By introducing the C-3WD theoretical model into the cloud load error analysis process,this dissertation proposes the cloud load prediction technology based on the C-3WD model under the perception of energy consumption.Firstly,the cloud resource demand is forecasted,and then the predicted error sequence is expressed as the error interval by using the interval set.Secondly,the idea of C-3WD is introduced,and the error sequence is divided into stable,jitter and volatility error intervals,and then the error interval is corrected by the corresponding prediction method,so as to enhance the prediction performance,improve resource utilization and effectively reduce energy consumption.