计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (6): 350-357.doi: 10.11896/jsjkx.250300164

• 计算机网络 • 上一篇    下一篇

基于Barrier的网络存储保序方法

韩磊1, 李谋兴1, 吴争1, 樊卫北1, 钱小燕2   

  1. 1 南京邮电大学计算机学院 南京 210000
    2 南京航空航天大学民航学院 南京 210000
  • 收稿日期:2025-03-31 修回日期:2025-06-10 出版日期:2026-06-15 发布日期:2026-06-09
  • 通讯作者: 李谋兴(1023040802@njupt.edu.cn)
  • 作者简介:(hanlei@njupt.edu.cn)

Barrier-based Network Storage Ordering Method

HAN Lei1, LI Mouxing1, WU Zheng1, FAN Weibei1, QIAN Xiaoyan2   

  1. 1 College of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210000,China
    2 College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 210000,China
  • Received:2025-03-31 Revised:2025-06-10 Published:2026-06-15 Online:2026-06-09
  • About author:HAN Lei,born in 1979,professor,se-nior engineer.His main research in-terests include data center architecture and networks,specifically protocols,algorithms,hardware acceleration and network-compute convergence.
    LI Mouxing,born in 2001,postgra-duate.His main research interests include computer network and distributed storage.

PDF (PC)

67

摘要： 在传统数据交换模型中,网络延迟和操作系统调度等因素可能导致数据传输和处理出现乱序,从而导致大量延迟,以及存储性能降低。为解决这一问题,提出一种基于Barrier机制的存储系统数据写入顺序控制方法和改进后的融合网络传输特性M/G/1队列模型。该方法利用特定的Barrier命令确保数据写入操作的顺序性,并基于排队论中改进后的融合网络传输特性M/G/1队列模型量化存储过程,提高系统可靠性和一致性;同时,引入了请求保序机制,进一步保证数据传输的顺序。实验表明,所提保序方法在保证数据一致性和顺序性的同时,相较于传统方法(部署于Linux的原生存储系统),在较差的网络环境下,能够平均提升25%的存储效率,在普通的网络环境中能够最高提升8%的存储效率。

关键词: 分布式存储, 屏障命令, 保序方法, M/G/1排队论, I/O效率

Abstract: In traditional data exchange models,network latency and operating system scheduling may lead to disorder in data transmission and processing,resulting in significant delays and degraded storage performance.To address this issue,this study proposes a Barrier mechanism-based sequential write control method for storage systems and an enhanced M/G/1 queue model incorporating network transmission characteristics.The method ensures write operation orderliness through dedicated Barrier commands and quantifies storage processes using the improved M/G/1 queue model from queuing theory,thereby enhancing system reliability and consistency.Additionally,a data packet ordering mechanism is introduced to guarantee transmission sequence.Experimental results demonstrate that compared to conventional approaches(implemented in native Linux storage systems),the proposed method improves storage efficiency by an average of 25% under poor network conditions and up to 8% in normal network environments,while maintaining data consistency and sequential integrity.

Key words: Distributed storage, Barrier command, Sequencing methods, M/G/1 queue theory, I/O efficiency

中图分类号: 

  • TP393
[1]VAN STEEN M,TANENBAUM A S.A brief introduction to distributed systems [J].Computing,2016,98(10):967-1009.
[2]CASAS I,TAHERI J.Analysis of a network IO bottleneck in big data environments based on docker containers [J].Big Data Research,2016,3:24-28.
[3]WON Y,JUNG J,CHOI G,et al.Barrier-enabled IO stack for flash storage [C]//Proceedings of the 16th USENIX Conference on File and Storage Technologies.Oakland:USENIX Association,2018:211-226.
[4]BACCELLI F,BRÉHIER A.Queueing Theory and its Applica-tions(2nd ed)[M].New York:Wiley,2015.
[5]JU R,WANG W,LI J,et al.On building a low latency network for future internet services [C]//GLOBECOM 2017－IEEE Global Communications Conference.IEEE,2017:1-6.
[6]ARPACI-DUSSEAU A C,ARPACI-DUSSEAU R H.Towards transparent CPU scheduling [R].Amherst:University of Wisconsin-Madison,2011.
[7]TAN Z,LIN S J,HAN Y S,et al.Data integrity check in distributed storage systems [C]//IEEE International Symposium on Information Theory.IEEE,2022:2397-2402.
[8]PATTERSON R H,GIBSON G,STODOLSKY D,et al.In-formed prefetching and caching[C]//Proceedings of the 15th ACM Symposium on Operating System Principles.ACM,1995:79-95.
[9]GUO H.A message ordering method,network node and storage medium:CN109104446A [P].2018-12-28.
[10]MENG X R.Method and device for request processing orderpreservation in distributed storage protocol:CN107277128A [P].2017-10-20.
[11]ALIBABA(China) Co.,Ltd.Method and system for data operation sequencing,method for data operation submission:CN116737680A [P].2023-09-12.
[12]AL-ASSAF M M,JIANG X,QIN X,et al.Informed prefetching for distributed multilevel storage systems [J].Journal of Signal Processing Systems,2018,90(4):619-640.
[13] LIAO J,CHEN S.Optimization of reading data via classifiedblock access patterns in file systems [J].IEEE Access,2016,4:9421-9427.
[14]ZENG S,LIAO X,GUO H,et al.Volley:Accelerating write-read orders in disaggregated storage[C]//Proceedings of the Nineteenth European Conference on Computer Systems.ACM,2024:657-673.
[15]SAKOV M,POURMEIDANI S,JONES M D,et al.HPC I/O throughput bottleneck analysis with explainable local models [C]//SC20:International Conference for High Performance Computing,Networking,Storage and Analysis.IEEE,2020:1-13.
[16]BEZ J L,BAUTISTA-GOMEZ L,MELLO R F,et al.I/O bottleneck detection and tuning:connecting the dots using interactive log analysis [C]//IEEE/ACM Sixth International Parallel Data Systems Workshop.IEEE,2021:15-22.
[17]ZHENG F,GAO C,ZHU G.Asymptotic stability of M/G/1queuing system [J].Journal of Applied Functional Analysis,2011,13(2):156-162.
[18]LITTLE J D C.A Proof of the Queuing Formula:L＝λW [J].Operations Research,1961,9(3):383-387.
[19]GUPTA R,ZHANG Y,PATEL S.Network-aware queuing theory for distributed storage optimization [J].IEEE Transactions on Cloud Computing,2022,10(3):451-467.
[20]JOSHI K,GUPTA A,GONZÁLEZ J,et al.I/O Passthru:upstreaming a flexible and efficient I/O path in Linux[C]//Proceedings of the 22nd USENIX Conference on File and Storage Technologies.Santa Clara:USENIX Association,2024:107-122.
[21]DUNN W L,SHULTIS J K.Exploring Monte Carlo Methods [M].Amsterdam:Elsevier,2011.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发