Steven H. Low, Larry Peterson, and Limin Wang, "Understanding TCP Vegas: A duality model," in Proceedings of SIGMETRICS, June 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Vegas falls back to Reno, and (c) Vegas utilizes the network more efficiently than Reno and avoids the bias of Reno against flows with long propagation delays. Mo et al. 21] also used a fluid approximation and also conclude that Vegas throughput isn t dependent on propagation delay. Low et al. [18] model Vegas as a distributed optimization algorithm. They show that Vegas uses queueing delay as a congestion measure and verify all the above findings. Using a duality model in [16] Low finds that Vegas achieves proportional fairness and that when Vegas and Reno flows share a common network, ....

....perform equally well for the entire range of network configurations, with maximum error around 20 . 4.4 Fairness of TCP Vegas The two main findings from previous investigations of TCP Vegas fairness, reviewed in 2. 2, are: 1) Vegas removes Reno s bias against flows with large propagation delays [4, 18, 21] , and (2) Vegas creates persistent congestion which causes unfair distribution of the throughput due to inaccurate measures of baseRTT [5, 18] However, these issues have primarily been analyzed in environments where the Vegas sources experience no loss. Our analytic models suggest that Vegas has ....

[Article contains additional citation context not shown here]

S. H. Low, L. L. Peterson, and L. Wang. Understanding TCP Vegas: A duality model. In SIGMETRICS/Performance 01, pages 226--235, Cambridge, MA, June 2001.

....and with 1 exogenous losses (right) 6. Related Work This paper is inspired by the need to merge two orthogonal research directions within the networking community. Control theoretic Analysis: On one hand, marshaling techniques from control and optimization theory has been a fruitful direction [4, 11, 12, 21, 16, 17, 8, 19, 18]. In particular, studies in [20, 11] investigated the stability regions for TCP over RED. Katabi et al.: 15] develops a controller for stable and efficient congestion control. Kelly et al.: 16] model TCP AQM as an optimization problem where they maximize the aggregate resource utility. However, ....

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A Duality Model. In Proceedings of ACM SIGMETRICS 2001.

....our work is a step toward enhancing transient measures like aggressiveness by exploring the design space between window based memory less control schemes and equation based schemes that make use of longer history. Much of the literature has focused on the modeling of TCP congestion control [24, 25, 26, 27, 28, 19, 29]. Ott et al.: showed that if packet losses are independent with small probability p, the average window size and long term throughput are of the order of 1= p. Lakshman et al.: 26] studied the properties of TCP in a regime where the bandwidth delay product is high and losses are random. In [27] ....

....model that also captures the e ect of TCP s timeout mechanism on throughput. Altman et al.: 24] analyze TCP throughput under a more general loss process which is assumed to be stationary. The model thus can account for any correlation and inter loss time distributions. Recently, Low et al.: [29] presented a duality model of TCP Vegas congestion control mechanism [30] 8 Conclusions We proposed a spectrum of TCP like window based congestion controls. Unlike memory less controls such as AIMD and binomial controls, our controls utilize history information. They are TCP friendly and ....

Steven H. Low, Larry Peterson, and Limin Wang, \Understanding TCP Vegas: A duality model," in Proceedings of SIGMETRICS, June 2001.

....represents a leap in our ability to model TCP Vegas sources. In particular two major limitations of other TCP models are overcome in our model: 1. Our model is the first model of TCP Vegas sources sending on off traffic; all previous work on modeling TCP Vegas only allowed bulk transfers [1, 2, 9, 12, 13, 16, 19]. Bulk transfer models are usually associated with FTP traffic, where This work was supported by NSF Career Grant CCR 0133077, NSF ITR Grant 99 167 ANI 0081396, Cisco Systems, Spinnaker Networks via Pittsburgh Digital Greenhouse Grant 01 1, an NSF Graduate Research Fellowship, and a project grant ....

....traffic, which inherently is made up of many short connections. 2. Our model is the first model of TCP Vegas that accurately predicts the operating point of the network in terms of throughput and loss rate for on off traffic. In particular, all but one previous work assumes loss free operation [1, 2, 9, 12, 13, 16]. Samios and Vernon [19] recently proposed the first analytical model to incorporate loss rate. Based on a renewal approach, they give a closed form formula for the throughput achieved by a bulk transfer of a single TCP Vegas sender as a function of the loss rate and the round trip time (RTT) of ....

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A duality model. In Proc. of ACM Sigmetrics, pages 226--235, Jun. 2001.

....and with 1 exogenous losses (right) 6. Related Work This paper is inspired by the need to merge two orthogonal research directions within the networking community. Control theoretic Analysis: On one hand, marshaling techniques from control and optimization theory has been a fruitful direction [4, 11, 12, 21, 16, 17, 8, 19, 18]. In particular, studies in [20, 11] investigated the stability regions for TCP over RED. Katabi ## ### [15] develops a controller for stable and efficient congestion control. Kelly ## ### [16] model TCP AQM as an optimization problem where they maximize the aggregate resource utility. However, ....

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A Duality Model. In Proceedings of ACM SIGMETRICS 2001.

....of a single sender as a function of the loss rate in the network. The model of TCP Vegas presented in this work overcomes two limitations of previous TCP models. First, our model is the first model of TCP Vegas sources sending on off traffic; all previous work only allowed bulk transfers [5, 6, 17, 21, 22, 25, 33]. Bulk transfer models are usually associated with FTP traffic, where a user sends a large amount of data without ever closing the connection. On off traffic provides a more general model, one that actually includes bulk transfer as a special case (when the on periods are very long) On off ....

....which inherently is made up of many short connections. Second, our model is the first model of TCP Vegas that accurately predicts the operating point of the network in terms of throughput and loss rate for on off traffic. Most of the prior work gives the throughput only under loss free operation [5, 6, 17, 21, 22, 25]. Samios and Vernon [33] recently proposed the first analytical model to incorporate loss rate. Based on a renewal approach, they give a closed form formula for the throughput achieved by a bulk transfer of a single TCP Vegas sender as a function of the loss rate and the round trip time (RTT) of ....

[Article contains additional citation context not shown here]

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A duality model. In Proceedings of ACM Sigmetrics, pages 226--235, June 2001.

....our work is a step toward enhancing transient measures like aggressiveness by exploring the design space between window based memory less control schemes and equation based schemes that make use of longer history. Much of the literature has focused on the modeling of TCP congestion control [25, 26, 27, 28, 29, 20, 30]. Ott et al. showed that if packet losses are independent with small probability p, the average window size and long term throughput are of the order of 1 # p. Lakshman et al. 27] studied the properties of TCP in a regime where the bandwidth delay product is high and losses are random. In [28] ....

....model that also captures the effect of TCP s timeout mechanism on throughput. Altman et al. 25] analyze TCP throughput under a more general loss process which is assumed to be stationary. The model thus can account for any correlation and inter loss time distributions. Recently, Low et al. [30] presented a duality model of TCP Vegas congestion control mechanism [31] 8 Conclusions We proposed a spectrum of TCP like window based congestion controls. Unlike memory less controls such as AIMD and binomial controls, our controls utilize history information. They are TCP friendly and ....

Steven H. Low, Larry Peterson, and Limin Wang, "Understanding TCP Vegas: A duality model," in Proceedings of ACM SIGMETRICS, June 2001.

....rate until another loss event is detected. This gives TCP its characteristic sawtooth transmission pattern, as it probes for bandwidth, experiences loss, backs off, and repeats the cycle. A more recent attempt to improve the congestion behaviour of TCP resulted in Vegas [Ahn95, Brakmo95, Low01] Rather than continually probing the network to see if it may increase its congestion window when it reaches steady state, it attempts to estimate the correct congestion window size. This is done by accurate estimation of the RTT using the assumption that the lowest RTT is the RTT that the ....

....worldwide network supporting an enormous number of users and services, from simple data transfer to more demanding soft real time multimedia applications. In large part the success of the Internet has been attributed to the simplicity of the service provided by IP, and the flexibility that this allows [Odlyzko00] However, this flexibility does come at a price precisely because IP is so simple, it provides very little support for more demanding applications. For example, congestion control had to be implemented in TCP after serious problems with congestion in the Internet arose [Jacobson88] and the ....

[Article contains additional citation context not shown here]

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A Duality Model. In Proceedings of ACM SIGMETRICS,

....in th router s bu#er. Weth deriveth th5 6R75O of TCP connections in steady state, and discussh w to 90 VO430: NO JANUARY 2002 configure control parameters of th window based flow control mechR7KS for achR7HH7 better fairness among TCP connections in th h755 ORxHK5O network. Note tht in [7],anoth approach hp been taken to derive th th65H Rxz of TCP Vegas.AlthR76 th thhR76 put obtained inth7 paper is identical toth of [7] our analysisha th following advantages: 1) an iterative computation is not necessary to obtainth thinR 5K and (2) th th67 Rxz of th TCP ....

....control parameters of th window based flow control mechR7KS for achR7HH7 better fairness among TCP connections in th h755 ORxHK5O network. Note tht in [7] anoth approach hp been taken to derive th th65H Rxz of TCP Vegas.AlthR76 th thhR76 put obtained inth7 paper is identical toth of [7], our analysisha th following advantages: 1) an iterative computation is not necessary to obtainth thinR 5K and (2) th th67 Rxz of th TCP connection can be derived algebraically forrathH simple network models. Next, we deriveth stability condition of th window based flow control ....

S. Low, L. Peterson, and L. Wang, "Understanding TCP Vegas: A duality model," Proc. ACM SIGMETRICS 2001.

No context found.

Steven H. Low, Larry Peterson, and Limin Wang, "Understanding TCP Vegas: A duality model," in Proceedings of SIGMETRICS, June 2001.

No context found.

Low, S., L. Peterson and L. Wang. (2001) Understanding TCP Vegas: a duality model. In: Proc. of ACM SIGMETRICS, 226-235.

No context found.

L. Peterson S. Low and L. Wang. Understanding TCP Vegas: A Duality Model. In Proceedings of the ACM SIGMETRICS 2001 Conference, pages 226--235, June 2001.

No context found.

S. Low, L. Peterson, L. Wang, Understanding TCP Vegas: a duality model, in: Proceedings of ACM SIGMETRICS, June 2001, pp. 226--235.

No context found.

S. Low, L. Peterson, L. Wang, Understanding TCPVegas: a duality model, Proceedings of ACM SIGMETRICS 2001.

No context found.

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A duality model. In Proceedings of ACM Sigmetrics, pages 226--235, June 2001.

No context found.

S. H. Low, L. L. Peterson and L. Wang, "Understanding TCP Vegas: A duality model," Proceedings os ACM SIGMETRICS, pp. 226-235, Boston, MA, June 2001.

No context found.

L. W. S. H. Low, Larry Peterson, "Understanding TCP Vegas: A duality model," june 2001.

No context found.

S. Low, L. Peterson, and L. Wang, "Understanding TCP vegas: a Duality Model," Journal of the ACM, vol. 49, no. 2, March 2002.

No context found.

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A Duality Model. In ACM SIGMETRICS 2001.

No context found.

L. Peterson S. Low and L. Wang. Understanding TCP Vegas: A Duality Model. In Proceedings of the ACM SIGMETRICS 2001 Conference, pages 226--235, June 2001.

No context found.

S. Low, L. Peterson, and L. Wang. Understanding TCP Vegas: A Duality Model. In ACM SIGMETRICS 2001.

No context found.

S. Low, L. Peterson, and L. Wang, "Understanding TCP Vegas: a Duality Model," ACM SIGMETRICS, 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC