Ceph Cluster Based Energy Consumption Management Strategy Study In Cloud Platform

Cloud computing and big data technology have developed rapidly in recent years.From smart healthcare to smart cities,from precision poverty alleviation to precision industry,from government offices to personalized recommendation systems,all of these are showing the changes that cloud computing and big data brought to people's lives.However,at the same time,the huge energy consumption of data centers has caused widespread concern.How to reduce the energy consumption of data center has become a hot topic in the industry and academia.Firstly,the research background,significance,research status and related basic knowledge of cloud platform energy consumption management are introduced;secondly,aiming at the shortcomings of CRUSH algorithm in energy saving,an energy consumption optimization model based on Ceph clusters is established and a data-place policy for energy saving based on this model is proposed.Ceph cluster can be divided into Operating zone and Standby zone under this strategy,and Sequential storage and Random storage are combined while considering the Service-Level Agreement(SLA)of users and the performance of Ceph cluster,which can improve the Ceph cluster from the perspective of Energey-saving-ability;Besides,in order to further save the energy consumption of Ceph cluster,a energy saving model--Docker cloud resource scheduling based on the improved Ceph cluster is established and a energy-saving cloud resource scheduling strategy,including deployment algorithm and application migration algorithm,based on this model is proposed,which combines the characteristics of Ceph cluster and data volumes of Docker container.The deployment algorithm and migration algorithm are used to improve the Docker Swarm orchestrator and solve the problem that container volumes cannot be migrated across hosts during application migration,respectively;finally,a series of experiments are desighed to verify the effectiveness and feasibility of the two proposed strategies.The experimental results show that the data-place policy can reduce the energy consumption of the data center by 14.3% under the condition of satisfying the users' SLA(Service-Level Agreement)requirement and ensuring cluster performance.Furthermore,compared with the original Docker Swarm scheduling strategy,the proposed cloud resources scheduling strategy performs a more fine-grained partitioning of cluster resources,which improves the utilization of cluster resources and further reduces the energy consumption of data center.