摘要
在军事应用场景下,各部队日常积累的海量数据具有部署分散、区域跨度大等特点,通过将跨域的原始数据汇总后再分析的方式面临着计算效率低、带宽压力大的挑战。本文在借鉴Hadoop、Spark等分布式计算框架的基础上,结合网格计算的思想提出了一种适用于军用网格环境的广域分布式数据处理框架,能够将各部队中分布松散的数据、计算资源聚合起来构成一个大规模计算系统,以提供在跨军网广域范围内的大数据分析处理能力。同时,模拟军事应用场景设计了仿真实验,实验结果证实本框架相对于传统计算方式在计算效率上有显著提高。
In military scene,the daily mass data accumulated by troops are characterized by dispersed deployment and large regional span. In order to collect original data from cross domain and analysis the big data,we are facing the challenges of low computational efficiency and high bandwidth pressure.Based on distributed computing,this paper proposed a wide area distributed data processing framework for military grid,which can aggregate distributed loose data resources and computing resources to form a large-scale computing system. It provides big data analysis and processing capabilities across the wide area of the military network. According to typical military application scenario,a simulation experiment is designed. The experiment results show that the efficiency of the framework is higher than that of traditional computing method.
作者
张智
江果
蒋鸣远
ZHANG Zhi;JIANG Guo;JIANG Ming-yuan(The 29th Research Institute of CETC,Chengdu 610036,China)
出处
《中国电子科学研究院学报》
北大核心
2019年第1期20-25,共6页
Journal of China Academy of Electronics and Information Technology
关键词
分布式计算
网格计算
广域
军用
任务调度
Distribute computing
Grid Computing
Wide Area
Military
Task scheduling
