摘要
物联网终端产生的大量数据给云数据中心造成巨大负担.边缘计算技术的出现分担了云端的部分数据处理,但是随着规模扩大,大量物联网终端数据处理对边缘计算通信性能提出了更高的要求.现有的一些通信处理手段都是通过减少数据量来提高整体性能,底层的通信效率问题并未得到有效解决.针对以上问题,本文提出一种能够在边缘计算节点通信中使用的基于DPDK的多端口并行通信机制,并加入改进的端口分发策略和反馈控制策略.实验表明,该机制不仅实现了基于数据链路层的多端口并行通信,还实现了DPDK底层通信进程和多个应用进程的高速数据交互以及根据底层通信控制模块合理分配带宽资源.
A large amount of data generated by Internet of Things terminals has imposed a huge burden on cloud data center.The emergence of edge computing technology shares some of the data processing in the cloud.With the expansion of the scale,A large number of Internet of Things terminal data processing put forward higher requirements for edge computing communication performance.Some existing communication processing methods improve the overall performance by reducing the amount of data,but the underlying communication efficiency problem has not been effectively solved.To solve the above problems,this paper proposes a multi-port parallel communication mechanism based on DPDK,which can be used in edge computing node communication,and adds an improved port distribution strategy and feedback control strategy.Experiments show that the mechanism not only realizes multi-port parallel communication based on data link layer,but also realizes high-speed data interaction between DPDK underlying communication process and multiple application processes,and can reasonably allocate bandwidth resources according to underlying communication control module.
作者
姜海粟
陈庆奎
JIANG Hai-su;CHEN Qing-kui(School of Optical-Electrical Computer Engineering,University of Shanghai for Science&Technology,Shanghai 200093,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2020年第5期1053-1058,共6页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61572325,60970012)资助
高等学校博士学科点专项科研博导基金项目(20113120110008)资助
上海重点科技攻关项目(14511107902,16DZ1203603)资助
上海市工程中心建设项目(GCZX14014)资助
上海智能家居大规模物联共性技术工程中心项目(GCZX14014)资助
上海市一流学科建设项目(XTKX2012)资助
沪江基金研究基地专项(C14001)资助
上海市重点项目(9DZ1208903)资助.
关键词
边缘计算
多端口并行
DPDK
进程交互
edge computing
multiport parallelism
DPDK
process interaction
