Differentiated Flow Transmission Control Protocol in Data Center Networks

doi:10.7652/xjtuxb201706020

您当前的位置：

首页 >

文章列表页 >

Differentiated Flow Transmission Control Protocol in Data Center Networks

更新时间：2025-07-09

- Differentiated Flow Transmission Control Protocol in Data Center Networks
- Vol. 51, Issue 6, Pages: 122-128+152(2017)
- 作者机构：
  
  重庆大学通信工程学院,重庆,400030
- 作者简介：
- 基金信息：
- DOI：10.7652/xjtuxb201706020
  CLC： TP393
- Online First：10 June 2017，
  
  Published：2017
- 稿件说明：
移动端阅览
Differentiated Flow Transmission Control Protocol in Data Center Networks[J]. 2017, 51(6): 122-128+152.
DOI：

Differentiated Flow Transmission Control Protocol in Data Center Networks[J]. 2017, 51(6): 122-128+152. DOI： 10.7652/xjtuxb201706020.

摘要

针对数据中心网络多对一通信流量产生TCP Incast拥塞导致吞吐量降低

以及小流易受大流影响难以满足应用截止时间要求等问题

提出了差分流传输控制协议(DFTCP)。DFTCP采用主动队列管理机制

利用显式拥塞通知机制传递拥塞信息

通过控制交换机队列长度来减少突发流分组的丢失以解决TCP Incast问题。DFTCP在终端服务器中对TCP流进行分类

当网络发生拥塞时

基于网络状态信息和流分类信息调节TCP拥塞窗口

通过对大流进行更大程度的拥塞退避从而减小其对网络中其他TCP流传输性能的影响

进而减少小流完成传输的时间

解决小流高延迟的问题。仿真实验表明:与传统TCP相比

DFTCP能够避免因TCP Incast拥塞导致的吞吐量崩溃; 与数据中心传输控制协议(DCTCP)相比

DFTCP能够减少小流传输完成的时间

同时DFTCP能够实现共享同一链路的多个大流快速收敛

保证公平性。

Abstract

Widely deployed applications in cloud computing with many-to-one communication pattern may cause TCP incast congestion problem

resulting in throughput collapse. Mice flows are easily influenced by elephant flows to miss the application deadline. To address these problems

we propose the differentiated flow transmission control protocol(DFTCP). The DFTCP adopts active queue management scheme and delivers the network congestion information through explicit congestion notifications. DFTCP reduces the number of discarding bursty small packets by controlling the queue length of switches to solve TCP incast problem. TCP flows are classified by DFTCP at the end-servers. When network congestion happens

DFTCP adjusts TCP congestion windows based on network states and classified flow information. Elephant flows have larger back-off time compared with mice flows. Simulation shows that DFTCP can effectively avoid TCP incast congestion compared with the traditional TCP

and it can reduce the completion time of mice flows compared with DCTCP. In addition

DFTCP can make multiple elephant flows converge quickly to guarantee the fairness of link sharing.

关键词

Keywords

references

ARMBRUST M, FOX A, GRIFFITH R, et al. Above the clouds: a Berkeley view of cloud computing: UCB/EECS-2009-28 [R/OL]. [2016-08-20].http:∥www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.html.

李丹, 陈贵海, 任丰原, 等. 数据中心网络的研究进展与趋势 [C]∥2011中国计算机科学技术年度报告: A卷. 北京: 机械工业出版社, 2012: 54-55.

ALIZADEH M, GRENNBERG A, MALTZ D, et al. Data center TCP(DCTCP)[J]. ACM SIGCOMM Computer Communication Review, 2010, 40(4): 63-74.

CHEN K, BAI W, ALIZADEH M. Scheduling mix-flows in commodity datacenters with Karuna [C]∥ACM SIGCOMM'16. New York, USA: ACM, 2016: 174-187.

JIANG C, LI D, XX M. LTTP: an LT-code based transport protocol for many-to-one communication in data centers [J]. IEEE Journal on Selected Areas in Communications, 2014, 32(1): 52-64.

GHEMAWAT S, GOBIOFF H, LEUNG S T. The Google file system [J]. ACM Sigops Operating Systems Review, 2003, 37(5): 29-43.

CHANG F, DEAN J, GHEMAWAT S, et al. Bigtable: a distributed storage system for structured data [J]. ACM Transactions on Computer Systems, 2008, 26(2): 4.

DEAN J, GHEMAWAT S. MapReduce: simplified data processing on large clusters [J]. Communications of the ACM, 2008, 51(1): 107-113.

PHANISHAYEE A, KREVAT E, VASUDEVAN V, et al. Measurement and analysis of TCP through-put collapse in cluster-based storage systems [C]∥USENIX FAST 2008. Berkeley, CA, USA: USENIX Association, 2008: 175-188.

CHEN Y, GRIFFITH R, LIU J, et al. Understanding TCP incast through put collapse in data center networks [C]∥ACM Workshop on Research on Enterprise Networking. New York, USA: ACM, 2009: 73-82.

VASUDEVAN V, PHANISHAYEE A, SHAH H, et al. Safe and effective fine-grained TCP retransmissions for datacenter communication [C]∥ACM SIGCOMM 2009. New York, USA: ACM, 2009: 303-314.

WU H, FENG Z, GUO C, et al. ICTCP: incast congestion control for TCP in data center networks [C]∥ACM Conference on Emerging Networking Experiment and Technology. New York, USA: ACM, 2010: 13-13.

JIANG C, LI D, XU M, et al. A coding-based approach to mitigate TCP incast in data center networks [C]∥International Conference on Distributed Computing Systems Workshops. Piscataway, NJ, USA: IEEE, 2012: 29-34.

GHOBADI M, YEGANEH S H, GANJALI Y. Rethinking end-to-end congestion control in software-defined networks [C]∥ACM Hotnets 2012. New York, USA: ACM, 2012: 1-6.

JOUET S, PEZAROS D P. Measurement-based TCP parameter tuning in cloud data centers [C]∥21st IEEE International Conference on Network Protocols(ICNP). New York, USA: ACM, 2013: 1-3.

JOUET S, PERKINS C, PEZAROS D. OTCP: SDN-managed congestion control for data center networks [C]∥Network Operations and Management Symposium. Piscataway, NJ, USA: IEEE, 2016: 1-9.

CURTIS A R, KIM W, YALAGANDULA P. Mahout: low-overhead datacenter traffic management using end-host-based elephant detection [C]∥IEEE INFOCOM 2011. Piscataway, NJ, USA: IEEE, 2011: 1629-1637.

AL-FARES M, RADHAKRISHNAN S, RAGHAVAN B, et al. Hedera: dynamic flow scheduling for data center networks [C]∥USENIX Conference on Networked Systems Design and Implementation. Berkeley, CA, USA: USENIX Association, 2010: 281-296.

WU H, JU J, LU G, et al. Tuning ECN for data center networks [C]∥8th International Conference on Emerging Networking Experiments and Technologies Archive. New York, USA: ACM, 2012: 25-36.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A Task Scheduling Algorithm for Large Graph Processing Cloud in Computing

Deployment and Scheduling of Virtual Machines in Cloud Computing: An “AHP” Approach

A Composite Document Model and Its Access Control Scheme in Cloud Computing

An Intra-Network Sharing Policy for Multi-Tenant Clouds

A Delay Scheduling Algorithm for Short Jobs

Related Author

苏森

刘一阳

黄庆佳

李健

代敏 1

王刚 1

林建材 1

桂小林 1

Related Institution

State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications

School of Electronics and Information Engineering, Xi'an Jiaotong University

Shaanxi Province Key Laboratory of Computer Network, Xi'an Jiaotong University

School of Computer Science and Technology, Xidian University

Faculty of Software, Fujian Normal University

⁰