一种基于ICN的网内协同存储机制  

An ICN⁃based in⁃network cooperative storage mechanism

作　　者：汪雨 韩锐[1,2] 党寿江 WANG Yu;HAN Rui;DANG Shoujiang(National Network New Media Engineering Research Center,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院声学研究所国家网络新媒体工程技术研究中心,北京100190 [2]中国科学院大学,北京100049

出　　处：《电子设计工程》2024年第20期1-5,共5页Electronic Design Engineering

基　　金：中国科学院战略性科技先导专项课题(XDC02070100)。

摘　　要：随着科学大数据的快速发展,海量科学数据对存储系统的高速持久化存储读写速率提出了新的挑战。信息中心网络(Information-Centric Networking,ICN)因其网内节点的存储能力,可支持数据的高速持久化存储。但现有的网内存储方法并没有联合考虑传输链路状态和节点负载,无法解决高通量应用场景下的存储IO峰值过载问题。针对上述问题,文中提出了一种基于ICN的协同存储机制,通过该机制中的节点选择和流量分割方法,充分利用转发路径上的多个存储节点分担存储任务。实验结果表明,相比较ECMP等流量分割机制,该方法能够更有效地提高系统整体的写入性能,实现存储负载均衡。With the rapid development of big data in science and research,the massive amount of scientific data has posed new challenges to read and write rates in the high-speed persistent storage of storage systems.The Information-Centric Networking(ICN)can support high-speed persistent storage of data due to the storage capacity of its in-network nodes.However,existing in-network storage methods do not jointly consider the transmission link state and node load,and cannot solve the storage IO peak overload problem in high throughput application scenarios.To address the above problems,An ICN-based cooperative storage mechanism is proposed in the paper,in which several storage nodes on the forwarding path are fully utilized to share the storage task by means of the node selection and traffic splitting methods in the mechanism.The experimental results show that compared to traffic partitioning mechanisms such as ECMP,this method is more effective in improving the overall write performance of the system and achieving storage load balancing.

关 键 词：网内存储 信息中心网络 负载均衡 流量分割 

分 类 号：TN919[电子电信—通信与信息系统]