基于容器调度技术的气象产品加工系统应用研究  

Application Research of Meteorological Product Processing System Based on Container Scheduling Technology

作　　者：秦运龙[1] 霍庆 何林 何文春[2] 高峰[2] 陈士旺 QIN Yunlong;HUO Qing;HE Lin;HE Wenchun;GAO Feng;CHEN Shiwang(Hubei Meteorological Information and Technical Support Center,Wuhan 430074,China;National Meteorological Information Center,Beijing 100081,China;Shaanxi Meteorological Information Center,Xi’an 710014,China)

机构地区：[1]湖北省气象信息与技术保障中心,湖北武汉430074 [2]国家气象信息中心,北京100081 [3]陕西省气象信息中心,陕西西安710014

出　　处：《微型电脑应用》2024年第7期23-26,31,共5页Microcomputer Applications

基　　金：国家发展改革委工程建设项目(发改农经[2019]1987号);湖北省气象局科技发展基金项目(2023Y07)。

摘　　要：随着研究型业务的快速发展,业务和科研人员急需一种在不降低计算效率、充分利用计算资源情况下支持可独立运行、高可用、高并发、个性化研发环境、快速部署移植的新技术支撑气象业务和科研发展。针对该需求对容器技术在产品加工系统(DPL)中的应用进行分析研究,研究表明Docker容器从系统逻辑层面可实现研发环境的个性化、作业执行的独立性和算法的高效快速移植,并在业务执行完后释放资源,保障资源的充分共享利用,同时采用基于容器的分布式架构Kubernetes技术实现容器的高并发和负载均衡调度,满足气象产品固定时间高并发执行的需求,且执行效率相对物理机集群并无降低,该技术可应用于国家级和省级气象产品加工系统(DPL)中常规气象产品业务和科研算法的调度运行。With the rapid development of research-oriented business,business and scientific research personnel urgently need a new technology to support the development of meteorological business and scientific research,which can run independently,highly available,highly concurrent,personalized R&D environment,rapid deployment and transplantation without reducing computing efficiency and making full use of computing resources.This paper analyzes the application of container technology in data processing line(DPL).The research shows that Docker container can realize the personalization of R&D environment and the independence of business execution from the system logic level,and release resources after business execution to ensure the full sharing and utilization of resources,At the same time,the container based distributed architecture Kubernetes technology is used to achieve high concurrency and load balancing scheduling of containers,which can meet the needs of high concurrency execution of meteorological products at fixed time,and the execution efficiency is no lower than that of physical machine cluster.This technology can be applied to the scheduling operation of routine meteorological products business and scientific research algorithm in national and provincial meteorological data processing line system.

关 键 词：容器 DOCKER 气象大数据云平台 镜像仓库 DPL Kubernetes 负载均衡 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]