Design And Implementation Of Cloud Service Platform For Coal Mine Intelligent Application

For a long time,coal as the basic energy in my country,strong and powerful,and strong development of my country's society and economy.At present,the main strategic goal of my country's coal mining industry is to vigorously promote intelligent construction and implement the upgrade transformation of coal mine industries.China Coal Energy Research Institute Co.,Ltd.(hereinafter referred to as "China Coal Energy Research Institute")Promote construction of coal mine intelligent service system.The so-called intelligent application is composed of coal mine data,intelligent analysis algorithm,and business applications.The implementation of business applications requires accurate intelligent analysis algorithm based,and the intelligent analysis algorithm is designed to depend on massive coal mine data.However,in coal mine production activities,different intelligent applications may be deployed in different regions,which brings challenges to intelligent applications,in addition to this,because coal mine business is particulartom,such as some intelligent applications Uninterrupted service,and some applications will encounter access to skyrocketing,which also provides higher requirements for the upgrading of intelligent applications and resource elastic scaling.In order to solve the above problems,this thesis starts from the three different types of users of China,coal mines,third-party software development companies,analyzes functional demand,combined with cloud computing,container,and container-proof Open source software is integrated,designed and implemented a coal mine intelligent application cloud service platform.The platform is mainly divided into image management modules,algorithm management modules,application management modules,storage volume management modules,and five modules of cluster management modules.Among them,in the mirror management module,a highly available Docker private mirror warehouse is implemented based on Docker Registry,which reduces the time delay from the official mirror warehouse download,and the user can make a mirror according to its own demand,upload to the private mirror warehouse;In the algorithm management module,an increase,deletion,modification,and viewing of the intelligent algorithm are implemented.In the application management module,the Kubernetes system using Google opens the unified management of intelligent applications,including intelligent applications and rapid deployment,application version Upgrade rollback,application resource elastic telescopic,and application deletion function;in the storage volume management module,use distributed file system Ceph to provide powerful persistent storage capabilities;in the cluster management module,use Prometheus + Grafana monitoring scheme The entire cluster itself,the cluster node,and the container running on the node perform safety monitoring,allowing the user to understand the resource usage of the cluster.In order to solve the problem of the Kubernetes resource scheduling algorithm logic single,resource utilization rate,this article is designed and implemented a resource scheduling policy based on label and priority,which allows users to customize POD priority,reboot from POD,priority Level and label three conditions seize resources,efficiently use physical resources.At the same time,this thesis is designed and implemented new resource elastic telescopic modules.The module not only considers the CPU utilization when performing resource elastic telescopic,but also incorporates memory utilization and user access,but also allows users to customize elastic telescopicism.Algorithm,more adapted to complex production environments.Finally,this thesis tests the main functional points and performance of the designimplemented cloud service platform,analyzes the scheduling distribution of intelligent applications in the cloud service platform and the response time of the user request application list page in concurrency,indicating that the analysis results show The resource scheduling strategy of this document can efficiently utilize the underlying resource in the scheduling of intelligent applications,and the cloud service platform can respond quickly to the user's interaction.