基于核心路由器的缓存需求分析
|
摘要
由于光纤传输和电子工艺发展速度的不均衡性,作为网络节点的核心路由器正在成为网络演进的瓶颈。路由器是一种存储转发设备,其内部的缓存容量是限制其性能的主要因素之一。
本文以核心路由器的缓存需求为研究目标,结合国家863重大课题“可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统”,在总结和分析了现行的核心路由器缓存设置法则及各种改进方案的基础上,寻求一种适合于实际网络需求的核心路由器缓存设置指导方案。针对现行法则及各种改进方案在功耗、成本、时延、丢包率和利用串等方面的不足,从丢包率、链路速率等多种性能需求入手,研究了核心路由器的缓存需求问题,并结合当前骨干网络链路利用率较低的实际情况,给出了一种更贴近于实际网络的核心路由器缓存设置法则。仿真结果表明,在牺牲一定链路利用率的条件下,新的缓存设置法则在丢包率性能方面明显优于现有的缓存设置方案,能够更好的满足核心路由器的缓存需求;最后利用该法则指导了NDSC自主研发的T比特路由器的缓存设置,有效降低了项目成本。本文的主要工作如下;
1.总结、分析和比较了当前核心路由器缓存设置法则及后续的一些改进方案,指出现行法则及各种改进方案在功耗、时延、丢包率、利用率和成本等方面的不足,给出了课题的研究方向,为课题的研究提供了参考依据和新的研究思路。
2.从当前骨干网络链路利用率较低且骨干链路的TCP业务流规模巨大的实际情况出发,综合考虑丢包率、链路速率等多种性能需求,通过对不同TCP业务类型的分析,给出了一个更贴近实际网络的核心路由器缓存设置法则。
3.利用网络仿真工具NS2,针对两种主要的TCP协议类型,对本文所提出的新的缓存设置法则进行了仿真验证,仿真结果为该法则提供了有力的佐证。
4.结合国家863项目“可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统”,给出了应用于T比特路由器的缓存设置方案,指导了工程实现,有效地降低了T比特路由器的成本。
Due to the technology development disproportion of optical fiber transmission and electronics, the core router as network core node has become the main bottleneck in the progress of modern networks. As Storage and forwarding equipment, the Router's buffer size is one of the most important factors which influence router's performance.
This paper aims to size buffers for core routers and is according to the requirements on one of key projects of the National High-Tech Research and Development Program of China (863 Program) named "The basic platform and experiment system for terabit scalable high performance IPv4/v6 routers". This dissertation tries to find out a new rule to sizing buffers for core routers, based on the analysis and comparison of router's buffer sizing rules in existence. Being dead against the deficiency such as consumption, cost, delay, loss rate, utilization, and considering the loss rate and utilization. This paper analyzes the buffer sizing for core routers. Considering the situation that the link utilization is very low in backbone networks, we propose a new rule of sizing buffers. The results of simulation show the new rule is better than the current ones. This new rule applied to Terabit Router successfully and reduced the cost of the project. The key contributions of this dissertation include:
1. This dissertation briefly compares the different results and proposals of routers buffer sizing, and point out the deficiencies in consumption, cost, delay, loss rate and utilization. Giving direction, supplying references and new methods for the dissertation.
2. In consideration of the low link utilization and the large volume of TCP flows in current backbone networks, we propose a more realistic rule for sizing the buffers of backbone routers, which take into account several performance metrics such as packet loss rate and link utilization simultaneity.
3. This paper verifies the conclusions by using ns2 network simulator and giving strong proof to the conclusions.
4. Based on the forementioned analysis and according to the requirements of 863 Terabit Router project and other factors, we present a scheme of buffer size, applying the scheme to Terabit Router successfully and reducing the cost of the project.
本文以核心路由器的缓存需求为研究目标,结合国家863重大课题“可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统”,在总结和分析了现行的核心路由器缓存设置法则及各种改进方案的基础上,寻求一种适合于实际网络需求的核心路由器缓存设置指导方案。针对现行法则及各种改进方案在功耗、成本、时延、丢包率和利用串等方面的不足,从丢包率、链路速率等多种性能需求入手,研究了核心路由器的缓存需求问题,并结合当前骨干网络链路利用率较低的实际情况,给出了一种更贴近于实际网络的核心路由器缓存设置法则。仿真结果表明,在牺牲一定链路利用率的条件下,新的缓存设置法则在丢包率性能方面明显优于现有的缓存设置方案,能够更好的满足核心路由器的缓存需求;最后利用该法则指导了NDSC自主研发的T比特路由器的缓存设置,有效降低了项目成本。本文的主要工作如下;
1.总结、分析和比较了当前核心路由器缓存设置法则及后续的一些改进方案,指出现行法则及各种改进方案在功耗、时延、丢包率、利用率和成本等方面的不足,给出了课题的研究方向,为课题的研究提供了参考依据和新的研究思路。
2.从当前骨干网络链路利用率较低且骨干链路的TCP业务流规模巨大的实际情况出发,综合考虑丢包率、链路速率等多种性能需求,通过对不同TCP业务类型的分析,给出了一个更贴近实际网络的核心路由器缓存设置法则。
3.利用网络仿真工具NS2,针对两种主要的TCP协议类型,对本文所提出的新的缓存设置法则进行了仿真验证,仿真结果为该法则提供了有力的佐证。
4.结合国家863项目“可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统”,给出了应用于T比特路由器的缓存设置方案,指导了工程实现,有效地降低了T比特路由器的成本。
Due to the technology development disproportion of optical fiber transmission and electronics, the core router as network core node has become the main bottleneck in the progress of modern networks. As Storage and forwarding equipment, the Router's buffer size is one of the most important factors which influence router's performance.
This paper aims to size buffers for core routers and is according to the requirements on one of key projects of the National High-Tech Research and Development Program of China (863 Program) named "The basic platform and experiment system for terabit scalable high performance IPv4/v6 routers". This dissertation tries to find out a new rule to sizing buffers for core routers, based on the analysis and comparison of router's buffer sizing rules in existence. Being dead against the deficiency such as consumption, cost, delay, loss rate, utilization, and considering the loss rate and utilization. This paper analyzes the buffer sizing for core routers. Considering the situation that the link utilization is very low in backbone networks, we propose a new rule of sizing buffers. The results of simulation show the new rule is better than the current ones. This new rule applied to Terabit Router successfully and reduced the cost of the project. The key contributions of this dissertation include:
1. This dissertation briefly compares the different results and proposals of routers buffer sizing, and point out the deficiencies in consumption, cost, delay, loss rate and utilization. Giving direction, supplying references and new methods for the dissertation.
2. In consideration of the low link utilization and the large volume of TCP flows in current backbone networks, we propose a more realistic rule for sizing the buffers of backbone routers, which take into account several performance metrics such as packet loss rate and link utilization simultaneity.
3. This paper verifies the conclusions by using ns2 network simulator and giving strong proof to the conclusions.
4. Based on the forementioned analysis and according to the requirements of 863 Terabit Router project and other factors, we present a scheme of buffer size, applying the scheme to Terabit Router successfully and reducing the cost of the project.
引文
[1]C.Villamizar,S.Cheng.High Performance TCP in ANSENT[J].ACM Computer Communications Review,1994,24(5);45-60.
[2]S.H.Low,F.Paganini,J.Wang,S.Adlakha,Doyle JC.Dynamics of TCP/RED and a scalable control[A].In;Proceedings of IEEE INFOCOM 2002[C].New York;IEEE Press,2002;239-248.
[3]S.Iyer,N.McKeown.Making parallel switches practical[A].In;Proceedings of IEEE INFOCOM 2001.Anchorage;IEEE Press,2001;1680-1687.
[4]R.Bush,D.Meyer.Some Internet Architectural Guidelines and Philosophy[S].RFC 3439,IETF,December 2002.
[5]J.Hennessy,D.Patterson.Computer Architecture[M].San Francisco;Morgan Kaufmarm Publishers Inc.,1996;126.
[6]I.Sundar,R.R.Kompella,N.McKeown.Analysis of a memory architecture for fast packet buffers[A].In;Proceedings of IEEE High Performance Switching and Routering[C].Dallas;IEEE Press,2001;368-373.
[7]汪斌强.可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统总体实施方案[R].郑州;国家数字交换系统工程技术研究中心,2005.
[8]W.S.Richard.TCP Illustrated,Volume 1-The Protocols[M].Boston;Addison-Wesley Professional,1994;105.
[9]G.Appenzeller.Sizing Router Buffers[D].Palo Alto;Stanford University,2004.
[10]S.Shenker,L.Zhang,D.Clark.Some observations on the dynamics of a congestion control algorithm[J].ACM Computer Communications Review,1990,20(5);30-39.
[11]L.Zhang,S.Shenker,C.David.Observations on the Dynamics of Congestion Control Algorithm;The Effects of Two-Way traffic[A].In;Proceedings of ACM SIGCOMM '91[C].Chapin AL;ACM Press,1991;133-148.
[12]S.Floyd.HighSpeed TCP for Large Congestion Windows[S].RFC 3649,IETF,December 2003.
[13]R.Gary,W.Wright,S.Richard.TCP Illustrated,Volume Ⅱ-Implementation[M].Boston;Addison-Wesley Professional,1994;147.
[14]J.Chuck,L.Fraleigh.Provisioning Internet Backbone Networks to Support Latency Sensitive Applications[D].Palo Alto;Stanford University,2002.
[15]G.Appenzeller,I.Keslassy,N.McKeown.Sizing Router Buffers[J].ACM Computer Communications Review,2004,34(4);281-292.
[16]B.Dhamdhere,C.Dovrolis.Open issue in router buffer sizing[J].ACM Sigcomm Computer Communication Review,2006,36(1);87-92.
[17]G.Raina,D.Wiscik.Buffer sizes for large multiplexers;TCP queueing theory and instability analysis[A].In;EuroNGI conference on Next Generation Internet Networks[C].Rome;IEEE Press,2005;173-180.
[18]G.Yashar,N.McKeown.Experiment study of router buffer sizing[R].Palo Alto;Stanford University,Department of Electrical Engineering,2006.
[19]M.Robert.Scalable TCP congestion control[A].In;Proceedings of IEEE INFOCOM 2000[C].Tel Aviv;IEEE Press,2000;1176-1183.
[20]H.Dhamdhere,C.Jiang,C.Dovrolis.Buffering sizing for Congested Internet Links[A].In;Proceedings of IEEE INFOCOM 2005[C].Miami;IEEE Press,2005;1072-1083.
[21]M.Enachescu,Y.Ganjali,A.Goel,N.McKeown,T.Roughgarden.Part Ⅲ;Router with very small buffers[J].ACM/SIGCOMM Computer Communication Review,2005,35(3);83-90.
[22]H.Park,E.F.Burmeister,S.Bjorlin,J.E.Bowers.40-gb/s optical buffer design and simulations[A].In;The 4th International Conference on Numerical Simulation of Optoelectronic Devices 2004[C].Santa Barbara;IEEE Press,2004;19-20.
[23]S.Aggarwal,C.Savage,T.Anderson.Understanding the performance of TCP pacing[A].In;Proceedings of the IEEE 1NFOCOM 2000[C].Tel-Aviv;IEEE Press,2000;1157-1165.
[24]G.V.Brugier,R.S.Stanojevic,D.J.Leith,R.N.Shorten.A critique of recently proposed buffer-sizing strategies[J].ACM SIGCOMM Computer Communication Review,2007,37(1);43-48.
[25]Y.Ganjali,N.McKeown.Update on Buffer sizing in Internet Routers[J].Computer Communications Review,2006,36(5);67-70.
[26]I.Matta,G.Liang.Differentiated Predictive Fair Service for TCP Flows[A].In;ICNP'2000[C].Osaka;IEEE Press,2000;49-58.
[27]A.Shaikh,R.Jennifer,S.Kang.Load-sensitive routing of long-lived IP flows[A].In;Proceedings of ACM SIGCOMM'99[C].Boston;IEEE Press,1999;215-226.
[28]R.Caceres,P.Danzig,S.Jamin,D.Mitzel.Characteristics of Wide-Area TCP/IP Conversations[A].In;Proceedings of ACM SIGCOMM'91[C].Zurich;IEEE Press,1991;101-112.
[29]K.Thompson,G.Miller,R.Wilder.Wide-area Internet Traffic Patterns and Characteristics[J].IEEE Network,1997,11(6);10-23.
[30]L.Breslau,P.Cao,L.Fan,G.Phillips,S.Shenker.Web Caching and Zipf-like Distributions;Evidence and Implications[A].In;Proceedings of INFOCOM '99.New York;IEEE Press,1999;126-134.
[31]C.Williamson.On Filter Effects in Web Caching Hierarchies[J].ACM Transactions on Internet Technology,2002,2(1);44-77.
[32]国家计算机网络应急技术处理协调中心.CNCERT/CC 2006年上半年网络安全报告[R].北京;国家计算机网络应急技术处理协调中心,2007.
[33]程光,龚俭.大规模高速网络流量测量研究[J].计算机工程与应用,2002,38(5);17-19.
[34]V.Jacobson.Congestion avoidance and control[J].ACM Computer Communication Review,1988,18(4);314-329.
[35]W.Stevens.TCP Slow Start,Congestion Avoidance,Fast Retransmit,and Fast Recovery Algorithms[S].RFC 2001,IETF,January 1997.
[36]S.Floyd,T.Henderson.The NewReno Modification to TCP's Fast Recovery Algorithm[S].RFC 2582,IETF,April 1999.
[37]M.Mathis,J.Mahdavi,S.Floyd,A.Romanow.TCP Selective Acknowledgment Options[S].RFC 2018,IETF,October 1996.
[38]A.S.Tanenbaum.Computer Networks 3rd ed[M].New Jersey;Prentice Hall,Inc.,1996;205.
[39]L.Brakmo,S.O.Malley,L.Peterson.TCP Vegas;New techniques for congestion detection and avoidance[A].In;Proceedings of the SIGCOMM'94[C].London;IEEE Press,1994;24-35.
[40]L.Brakmo,L.Peterson.TCP Vegas;End to End Congestion Avoidance on a Global Internet[J].IEEE Journal on Selected Areas in Communications,1995,13(8);1465-1480.
[41]T.Kelly.Scalable TCP;Improving Performance in Highspeed Wide Area Networks[J].ACM SIGCOMM Computer Communication Review,2003,33(2);83-91.
[42]C.Jin,D.X.Wei,S.H.Low.FAST TCP;Motivation,Architecture,Algorithms,Performance[A].In;Proceedings of IEEE INFOCOM'04[C].Hong Kong;IEEE Press,2004;2490-2501.
[43]J.Padhye,V.Firoiu,D.F.Towsley.Modeling TCP reno performance;a simple model and its empirical validation[J].IEEE/ACM Transactions on Networking,2000,8(2);133-45.
[44]J.Youngmi,A.Gilbert,N.McKeown.Congestion control and periodic behavior[A].In;Proceedings of IEEE LANMAN Workshop[C].Boulder;IEEE Press,2001;76-86.
[45]Z.Lixia,D.David,G.Clark.Oscillating behaviour of network traffic;A case study simulation[J].Internet-working;Research and Experience,1990,1(2);101-112.
[46]S.Floyd,V.Jacobson.Random early detection gateways for congestion avoidance[J].IEEE/ACM Transactions on Networking,1993,1(4);397-413.
[47]Q.Lili,Y.Zhang,K.Srinivasan.Understanding the performance of many TCP flows[J].Computer Networks,2001,37(4);277-306.
[48]G.Iannaccone,M.Martin,D.Christophe.Aggregate traffic performance with active queue mamagement and drop from tail[J].ACM SIGCOMM Computer Communication Review,2001,31(3);4-13.
[49]邱菡,李玉峰,邬江兴,王凯.基于网络层的核心路由器缓存容量分析[J].微电子学与计算机,2006,23(9);105-107.
[50]M.Mellia,H.Zhang.TCP model for short lived flows[J].IEEE Communications Letters,2002,6(2);85-87.
[51]S.Yilmaz,I.Matta.On class-based isolation of UDP,short-lived and long-lived TCP flows[A],In;Proceedings of International Workshop on Modeling,Analysis and Simulation of Computer and Telecommunications Systems[C].Cincinnati;IEEE Press,2001;415-422.
[52]J.Cao,W.S.Cleveland,D.Lin,D.X.Sun.Internet traffic tends toward poisson and independent as the load increase[A].In;Proceedings of the Nonlinear Estimation and Classification[C].New York;Springer-Verlag,2002;83-109.
[53]T.Karagiannis,M.Molle,M.Faloutsos,A.Broido.A nonstationary poisson view of internet traffic[A].In;Proceedings of IEEE INFOCOM2004[C].Hong Kong;IEEE Press,2004;1558-1569.
[54]G.Cheng,J.Gong.Traffic behavior analysis with Poisson sampling on high-speed network[A].In;International Conferences on Info-tech & Info-net[C].Beijing,IEEE Press,2001;158-163.
[55]M.Lee,E.J.Kim,M.Yousif.Security Enhancement in InfiniBand Architecture[A].In;Proceedings of the 19th IEEE International Parallel & Distributed Processing Symposium[C].Denver,IEEE Press,2005;105-113.
[56]N.Wisitpongphan.J.M.Peha.Effect of TCP on self-similarity of network traffic[A].In;Proceedings of 12th IEEE International Conference on Computer Communications and Networks[C].Baltimore;IEEE Press,2003;370-373.
[57]V.Paxson.S.Floyd.Wide area traffic;the failure of Poisson modeling[J].IEEE/ACM Transactions on Networking,1995,3(3);226-244.
[58]A.Feldmann,C.Anna.T.Gilbert,W.Willinger.Data networks as cascades;Investigating the multifractal nature of internet WAN traffic[A].In;Proceedings of ACM SIGCOMM'98[C],Vancouver;IEEE Press,1998;25-38.
[59]M.Garetto,L.Cigno,R.Meo,M.A.Marsan.Modeling short-lived TCP connections with open multiclass queuing network[A].In;Proceedings of the 7th IFIP/IEEE International Workshop on Protocols for High Speed Networks[C],London;IEEE Press,2002;100-116.
[60]陈尚兵,王彬,钱积新.TCP拥塞控制综述[J].计算机科学,2002,29(5);32-35.
[61]W.Ronald,H.Wolff.Stochastic Modeling and the Theory of queues[M].New Jersey;Prentice Hall,1989;279.
[62]G.Michele,D.Towsley.Modeling,simulation and measurements of queueing delay under long-tail internet traffic[J].ACM SIGMETRICS Performance Evaluation Review,2003,31(1);47-57.
[63]F.P.Kelly.Stochastic Networks Theory and Applications[M].Oxford;Oxford University Press,1996;75-89.
[2]S.H.Low,F.Paganini,J.Wang,S.Adlakha,Doyle JC.Dynamics of TCP/RED and a scalable control[A].In;Proceedings of IEEE INFOCOM 2002[C].New York;IEEE Press,2002;239-248.
[3]S.Iyer,N.McKeown.Making parallel switches practical[A].In;Proceedings of IEEE INFOCOM 2001.Anchorage;IEEE Press,2001;1680-1687.
[4]R.Bush,D.Meyer.Some Internet Architectural Guidelines and Philosophy[S].RFC 3439,IETF,December 2002.
[5]J.Hennessy,D.Patterson.Computer Architecture[M].San Francisco;Morgan Kaufmarm Publishers Inc.,1996;126.
[6]I.Sundar,R.R.Kompella,N.McKeown.Analysis of a memory architecture for fast packet buffers[A].In;Proceedings of IEEE High Performance Switching and Routering[C].Dallas;IEEE Press,2001;368-373.
[7]汪斌强.可扩展到T比特的高性能IPv4/v6路由器基础平台及实验系统总体实施方案[R].郑州;国家数字交换系统工程技术研究中心,2005.
[8]W.S.Richard.TCP Illustrated,Volume 1-The Protocols[M].Boston;Addison-Wesley Professional,1994;105.
[9]G.Appenzeller.Sizing Router Buffers[D].Palo Alto;Stanford University,2004.
[10]S.Shenker,L.Zhang,D.Clark.Some observations on the dynamics of a congestion control algorithm[J].ACM Computer Communications Review,1990,20(5);30-39.
[11]L.Zhang,S.Shenker,C.David.Observations on the Dynamics of Congestion Control Algorithm;The Effects of Two-Way traffic[A].In;Proceedings of ACM SIGCOMM '91[C].Chapin AL;ACM Press,1991;133-148.
[12]S.Floyd.HighSpeed TCP for Large Congestion Windows[S].RFC 3649,IETF,December 2003.
[13]R.Gary,W.Wright,S.Richard.TCP Illustrated,Volume Ⅱ-Implementation[M].Boston;Addison-Wesley Professional,1994;147.
[14]J.Chuck,L.Fraleigh.Provisioning Internet Backbone Networks to Support Latency Sensitive Applications[D].Palo Alto;Stanford University,2002.
[15]G.Appenzeller,I.Keslassy,N.McKeown.Sizing Router Buffers[J].ACM Computer Communications Review,2004,34(4);281-292.
[16]B.Dhamdhere,C.Dovrolis.Open issue in router buffer sizing[J].ACM Sigcomm Computer Communication Review,2006,36(1);87-92.
[17]G.Raina,D.Wiscik.Buffer sizes for large multiplexers;TCP queueing theory and instability analysis[A].In;EuroNGI conference on Next Generation Internet Networks[C].Rome;IEEE Press,2005;173-180.
[18]G.Yashar,N.McKeown.Experiment study of router buffer sizing[R].Palo Alto;Stanford University,Department of Electrical Engineering,2006.
[19]M.Robert.Scalable TCP congestion control[A].In;Proceedings of IEEE INFOCOM 2000[C].Tel Aviv;IEEE Press,2000;1176-1183.
[20]H.Dhamdhere,C.Jiang,C.Dovrolis.Buffering sizing for Congested Internet Links[A].In;Proceedings of IEEE INFOCOM 2005[C].Miami;IEEE Press,2005;1072-1083.
[21]M.Enachescu,Y.Ganjali,A.Goel,N.McKeown,T.Roughgarden.Part Ⅲ;Router with very small buffers[J].ACM/SIGCOMM Computer Communication Review,2005,35(3);83-90.
[22]H.Park,E.F.Burmeister,S.Bjorlin,J.E.Bowers.40-gb/s optical buffer design and simulations[A].In;The 4th International Conference on Numerical Simulation of Optoelectronic Devices 2004[C].Santa Barbara;IEEE Press,2004;19-20.
[23]S.Aggarwal,C.Savage,T.Anderson.Understanding the performance of TCP pacing[A].In;Proceedings of the IEEE 1NFOCOM 2000[C].Tel-Aviv;IEEE Press,2000;1157-1165.
[24]G.V.Brugier,R.S.Stanojevic,D.J.Leith,R.N.Shorten.A critique of recently proposed buffer-sizing strategies[J].ACM SIGCOMM Computer Communication Review,2007,37(1);43-48.
[25]Y.Ganjali,N.McKeown.Update on Buffer sizing in Internet Routers[J].Computer Communications Review,2006,36(5);67-70.
[26]I.Matta,G.Liang.Differentiated Predictive Fair Service for TCP Flows[A].In;ICNP'2000[C].Osaka;IEEE Press,2000;49-58.
[27]A.Shaikh,R.Jennifer,S.Kang.Load-sensitive routing of long-lived IP flows[A].In;Proceedings of ACM SIGCOMM'99[C].Boston;IEEE Press,1999;215-226.
[28]R.Caceres,P.Danzig,S.Jamin,D.Mitzel.Characteristics of Wide-Area TCP/IP Conversations[A].In;Proceedings of ACM SIGCOMM'91[C].Zurich;IEEE Press,1991;101-112.
[29]K.Thompson,G.Miller,R.Wilder.Wide-area Internet Traffic Patterns and Characteristics[J].IEEE Network,1997,11(6);10-23.
[30]L.Breslau,P.Cao,L.Fan,G.Phillips,S.Shenker.Web Caching and Zipf-like Distributions;Evidence and Implications[A].In;Proceedings of INFOCOM '99.New York;IEEE Press,1999;126-134.
[31]C.Williamson.On Filter Effects in Web Caching Hierarchies[J].ACM Transactions on Internet Technology,2002,2(1);44-77.
[32]国家计算机网络应急技术处理协调中心.CNCERT/CC 2006年上半年网络安全报告[R].北京;国家计算机网络应急技术处理协调中心,2007.
[33]程光,龚俭.大规模高速网络流量测量研究[J].计算机工程与应用,2002,38(5);17-19.
[34]V.Jacobson.Congestion avoidance and control[J].ACM Computer Communication Review,1988,18(4);314-329.
[35]W.Stevens.TCP Slow Start,Congestion Avoidance,Fast Retransmit,and Fast Recovery Algorithms[S].RFC 2001,IETF,January 1997.
[36]S.Floyd,T.Henderson.The NewReno Modification to TCP's Fast Recovery Algorithm[S].RFC 2582,IETF,April 1999.
[37]M.Mathis,J.Mahdavi,S.Floyd,A.Romanow.TCP Selective Acknowledgment Options[S].RFC 2018,IETF,October 1996.
[38]A.S.Tanenbaum.Computer Networks 3rd ed[M].New Jersey;Prentice Hall,Inc.,1996;205.
[39]L.Brakmo,S.O.Malley,L.Peterson.TCP Vegas;New techniques for congestion detection and avoidance[A].In;Proceedings of the SIGCOMM'94[C].London;IEEE Press,1994;24-35.
[40]L.Brakmo,L.Peterson.TCP Vegas;End to End Congestion Avoidance on a Global Internet[J].IEEE Journal on Selected Areas in Communications,1995,13(8);1465-1480.
[41]T.Kelly.Scalable TCP;Improving Performance in Highspeed Wide Area Networks[J].ACM SIGCOMM Computer Communication Review,2003,33(2);83-91.
[42]C.Jin,D.X.Wei,S.H.Low.FAST TCP;Motivation,Architecture,Algorithms,Performance[A].In;Proceedings of IEEE INFOCOM'04[C].Hong Kong;IEEE Press,2004;2490-2501.
[43]J.Padhye,V.Firoiu,D.F.Towsley.Modeling TCP reno performance;a simple model and its empirical validation[J].IEEE/ACM Transactions on Networking,2000,8(2);133-45.
[44]J.Youngmi,A.Gilbert,N.McKeown.Congestion control and periodic behavior[A].In;Proceedings of IEEE LANMAN Workshop[C].Boulder;IEEE Press,2001;76-86.
[45]Z.Lixia,D.David,G.Clark.Oscillating behaviour of network traffic;A case study simulation[J].Internet-working;Research and Experience,1990,1(2);101-112.
[46]S.Floyd,V.Jacobson.Random early detection gateways for congestion avoidance[J].IEEE/ACM Transactions on Networking,1993,1(4);397-413.
[47]Q.Lili,Y.Zhang,K.Srinivasan.Understanding the performance of many TCP flows[J].Computer Networks,2001,37(4);277-306.
[48]G.Iannaccone,M.Martin,D.Christophe.Aggregate traffic performance with active queue mamagement and drop from tail[J].ACM SIGCOMM Computer Communication Review,2001,31(3);4-13.
[49]邱菡,李玉峰,邬江兴,王凯.基于网络层的核心路由器缓存容量分析[J].微电子学与计算机,2006,23(9);105-107.
[50]M.Mellia,H.Zhang.TCP model for short lived flows[J].IEEE Communications Letters,2002,6(2);85-87.
[51]S.Yilmaz,I.Matta.On class-based isolation of UDP,short-lived and long-lived TCP flows[A],In;Proceedings of International Workshop on Modeling,Analysis and Simulation of Computer and Telecommunications Systems[C].Cincinnati;IEEE Press,2001;415-422.
[52]J.Cao,W.S.Cleveland,D.Lin,D.X.Sun.Internet traffic tends toward poisson and independent as the load increase[A].In;Proceedings of the Nonlinear Estimation and Classification[C].New York;Springer-Verlag,2002;83-109.
[53]T.Karagiannis,M.Molle,M.Faloutsos,A.Broido.A nonstationary poisson view of internet traffic[A].In;Proceedings of IEEE INFOCOM2004[C].Hong Kong;IEEE Press,2004;1558-1569.
[54]G.Cheng,J.Gong.Traffic behavior analysis with Poisson sampling on high-speed network[A].In;International Conferences on Info-tech & Info-net[C].Beijing,IEEE Press,2001;158-163.
[55]M.Lee,E.J.Kim,M.Yousif.Security Enhancement in InfiniBand Architecture[A].In;Proceedings of the 19th IEEE International Parallel & Distributed Processing Symposium[C].Denver,IEEE Press,2005;105-113.
[56]N.Wisitpongphan.J.M.Peha.Effect of TCP on self-similarity of network traffic[A].In;Proceedings of 12th IEEE International Conference on Computer Communications and Networks[C].Baltimore;IEEE Press,2003;370-373.
[57]V.Paxson.S.Floyd.Wide area traffic;the failure of Poisson modeling[J].IEEE/ACM Transactions on Networking,1995,3(3);226-244.
[58]A.Feldmann,C.Anna.T.Gilbert,W.Willinger.Data networks as cascades;Investigating the multifractal nature of internet WAN traffic[A].In;Proceedings of ACM SIGCOMM'98[C],Vancouver;IEEE Press,1998;25-38.
[59]M.Garetto,L.Cigno,R.Meo,M.A.Marsan.Modeling short-lived TCP connections with open multiclass queuing network[A].In;Proceedings of the 7th IFIP/IEEE International Workshop on Protocols for High Speed Networks[C],London;IEEE Press,2002;100-116.
[60]陈尚兵,王彬,钱积新.TCP拥塞控制综述[J].计算机科学,2002,29(5);32-35.
[61]W.Ronald,H.Wolff.Stochastic Modeling and the Theory of queues[M].New Jersey;Prentice Hall,1989;279.
[62]G.Michele,D.Towsley.Modeling,simulation and measurements of queueing delay under long-tail internet traffic[J].ACM SIGMETRICS Performance Evaluation Review,2003,31(1);47-57.
[63]F.P.Kelly.Stochastic Networks Theory and Applications[M].Oxford;Oxford University Press,1996;75-89.