摘要
近年来多载荷、多分辨率的遥感影像数量快速增加,各行业对于遥感产品多样化、高时效性的需求日益增长,传统的生产方式已无法满足需求,亟需构建自动化、批量化的生产模式。为此,该文提出一种基于云原生技术架构的遥感大数据在线处理流程编排与算法调度技术,以容器为算法模型封装载体,以Kubernetes作为底层运行平台,构建流程驱动和资源调度双层架构,支持不同遥感算法的快速集成和用户自定义的生产任务全流程编排,搭建了具备可视化流程设计管理、标准化算法扩展、集群计算运行管理功能的遥感数据处理云服务平台,为遥感大数据批量处理与生产提供高效的解决方案。
In recent years,as the number of remote sensing images with various platforms and multi⁃resolution is increasing rapidly,and the demand for diversity and timely effectiveness of remote sensing products in different industries is growing,the traditional production mode has been unable to meet the demand.It is urgent to build an automated and batch production model.Therefore,a process orchestration and algorithm scheduling technology for remote sensing data online processing,which based on cloud native architecture,is proposed.The container is used as the encapsulation carrier of algorithm,and Kubernetes is used as the underlying platform.Process driven and resource scheduling two⁃layer architecture is established,in order to support the rapid integration of different remote sensing algorithms and user⁃defined production task with whole flow orchestration.Furthermore,a cloud service platform for remote sensing data processing has been built,which has functions of visual process design and management,standardized algorithm expansion,and cluster computing operation management,to provide efficient solutions for the batch processing and production of remote sensing big data.
作者
廖芳芳
雷雨桐
苏永恒
刘富乾
管占磊
吴聆益
LIAO Fangfang;LEI Yutong;SU Yongheng;LIU Fuqian;GUAN Zhanlei;WU Lingyi(PIESAT Information Technology Co.,Ltd.,Beijing 100195,China;Shanghai PIESAT Space Network Technology Co.,Ltd.,Shanghai 201702,China)
出处
《电子设计工程》
2022年第18期165-170,共6页
Electronic Design Engineering
基金
高分专项长三角区域军民融合应用示范项目(20511108000)。
关键词
云原生
遥感大数据
流程编排
算法调度
工作流引擎
并行计算
cloud native
remote sensing big data
process orchestration
algorithm scheduling
workflow engine
parallel computing
