Research On Task Scheduing Of Cloud Platform Based On Cloud User Application Evaluation

With the popularity of cloud computing, more and more users use this on-demand service model to solve business problems. Cloud computing service providerhelp users manage the infrastructure resources and schedule tasks. Task schedulingproblem has been a hot issue of resource management. Task scheduling emerge newgoals in cloud computing environments. Cloud service providers often want to getmore economic benefits when scheduling tasks, which requires cloud platform toreduce the system operating costs and confirm tasks’ running time. Current cloudcomputing data center maintenance costs, especially energy costs, become thesecond largest cost of the cloud system cost merely after infrastructure costs. Thusreducing energy consumption has become a major goal of task scheduling in cloudcomputing.As cloud computing data centers continues to expand, the energy consumptionof data centers will also be an exponential growth rate, energy problems havegradually become a cloud computing data center to be solved by the key problem.While saving data center energy consumption is not only for cloud computingservice providers, but also for the protection of the environment, prevention of globalwarming is of great significance. Most of the cloud computing service providers usedynamic conditioning equipment’s operating power to save energy. However thismeasure will greatly affect the overall system performance. So we propose a taskscheduling method of cloud platform based on cloud user application evaluation,which does not affect system performance while reducing the total energyconsumption of the system, helping cloud computing service providers to obtainhigher profits.Firstly, we introduce the current development of cloud computing technology.We also describe the current energy-saving methods for cloud service providers toreduce system power consumption. However, these methods all have limitations anddeficiencies. Secondly, we propose a task scheduling strategy on cloud platformbased on cloud user application evaluation. It is divided into two parts, the first partis tasks’ resource and energy evaluation, the second part is the scheduling strategy.During the first part, we first map user tasks to an unified structure, and theprediction strategy is based on this structure. By diving user’s program into a seriesof action sequences, we predict the energy consumption. At the second part, we offer a balance of performance, power consumption, and user SLA task schedulingstrategy according the prediction result of the first part. We improve the current cloudscheduling system by server organizational structure, resource allocation scheme,tasks and task scheduling algorithm. Finally, we compare with existing cloudplatform scheduling scheme according to an experiment.