W. Feng, D. Kandlur, D. Saha, K. Shin, "A Self-Configuring RED Gateway", in Proc. of IEEE Infocom, New York, NY, 1999, pp.1320-1328.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... The dropping probability was chosen to be a deterministic function of the average queue as explained in [8] Several variants of the RED mechanism with di#erent refinements have also been developed, such as weighted RED (WRED) cf. e.g. 5] RED with an in out bit (RIO) 6] adaptive RED (ARED) [7], stabilized RED (SRED) 15] and flow RED (FRED) 14] The use of RED has also been proposed in the Assured Forwarding (AF) which is one of the di#erentiated services (Di#Serv) tra#c handling mechanisms. In fact, some of these mechanisms are today already being deployed in routers of the ....

W. Feng, D. Kandlur, D. Saha and K. Shin, A self-configuring RED gateway, Proceedings of INFOCOM '99, Eighteenth Annual Joint Conference of the IEEE Computer and Communications Societies, Vol. 3 (1999), pp 1320-1328.

....simulation [9] 10] 11] technology. To the best of our knowledge, our flexible approach is unique and the first of its kind as applied to IP network management. There have been proposed configurationsupport or adaptive protocol techniques for specific problems (eg: BGP [12] OSPF [13] RED [14]) We discuss these related works in later sections. The crucial component of the framework is an efficient parameter state space search algorithm. The desired algorithm is required to: a) be scalable to large dimensional parameter state spaces; b) find good solutions quickly; c) be robust to ....

.... than DropTail[35] Therefore, IETF has recommended RED as the single active buffer management for wide deployment in the Internet[36] However, the setting of RED parameters has proved to be highly sensitive to network scenarios and the performance of misconfigured RED may suffer significantly [14], 37] 38] Therefore, RED needs constant tuning to adapt to the prevailing network conditions. In view of this, it has been debated whether or not RED can achieve its claimed advantages[38] 39] 40] Based on simplified models, some general guidelines for setting RED parameters have been ....

[Article contains additional citation context not shown here]

Wu chang Feng, Dilip D. Kandlur, Debanjan Saha, and Kang G. Shi. A self-configuring RED gateway. Proceedings of IEEE Infocom 1999.

.... than DropTail[2] Therefore, IETF has recommended RED as the single active queue management for wide deployment in the Internet [3] However, the setting of RED parameters has proved to be highly sensitive to network scenarios and the performance of misconfigured RED may suffer significantly [4], 5] 6] In addition, since network is a dynamic system, RED needs constant tuning to adapt to current network conditions. In view of this, it has been debated whether or not RED can achieve its claimed advantages[6] 7] 8] This paper attempts to address the following questions: Given ....

....TCP is not yet clearly understood. Based on simplified models, some general guidelines for setting RED parameters have been proposed[2] 5] 9] Intuitive modifications on RED have also been proposed to automate the tuning of RED under varying network conditions by adjusting one of the parameters[4], 10] However, the effectiveness of these methods in complex network scenarios is still under investigation. Rather than relying on simplified models or intuition, this paper exploits the advantage of network simulation technique and formulates the optimal configuration of RED as a black box ....

[Article contains additional citation context not shown here]

Wu chang Feng, Dilip D. Kandlur, Debanjan Saha, and Kang G. Shi. A self-configuring RED gateway. Proceedings of IEEE Infocom 1999.

.... If the load is light, then the average queue size stays close to q min and the system tends to be unstable [6] In order to address this problem Floyd et al. 3] have proposed a dynamic version of RED mechanism, called ARED, which is based on the mechanism originally proposed by Feng et al. [1]. ARED dynamically updates p max so that the average queue size stays close to a target queue size q target [3] Once every # interval period ARED compares the average queue size to the target queue size. If the average queue size is larger than target queue size and p max 0.5, then p max is ....

W. Feng, D. Kandlur, D. Saha, and K. Shin. A self-configuring RED gateway. Proc. IEEE INFOCOM, 1999.

....rate) smoothing the response signal. It was soon realized that RED required tuning according to the expected number of flows and the RTTs seen by these flows, or, in other words, the aggressiveness of the flow aggregate. Many mechanisms were proposed to automagically adjust for this [9] 16] [17]. Still, the result is an often bursty queue length and frequent queue overruns. Similar approaches were used for ATM ABR [18] where a system analogous to TCP with ECN was proposed, but with the capability of signaling several levels of congestion. Plasser et al. 19] studied nonlinear drop ....

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proceedings of the 17th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM '99), pages 1320--1328, March 1999.

....average packet sizes in our two datasets falls within a range of 500 bytes suggesting that a static MPS could be derived for a particular output link. Another approach to the problem of choosing an MPS parameter is to derive it from the traffic flowing through the bottleneck dynamically (much like [FKSS99] suggests other RED parameters be derived) While this approach may yield a more accurate MPS setting there are details that need worked out (e.g. how to average packet sizes and over what timescales) In addition, a dynamic calculation costs additional CPU cycles. Without further study we cannot ....

Wu-Chang Feng, Dilip Kandlur, Debanjan Saha, and Kang Shin. A Self-Configuring RED Gateway. In IEEE InfoCom, 1999.

....On the other hand, TDRR maintains an exponentially weighted throughput average for each flow, allowing it to remember events much longer in the past than the queue time constant; this allows it to enforce fairness, even for small buffer sizes. 5. 3 Self Configuring RED Self configuring RED [16] is a proposal for an adaptive RED policy that can self parameterize given different congestion types. This policy is similar to Blue, where RED s dropping probability, o ( is decreased when the average queue size falls below n and increased when the average queue size ....

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A SelfConfiguring RED Gateway", in IEEE INFOCOM 1999.

....marked acknowledgment reaches the sender, the sender reduces its throughput, as if a loss had happened in the network. The emergence of ECN has stimulated research on appropriate marking algorithms that indicate congestion to TCP sources to avoid packet losses resulting from buffer overflows [4, 14, 15, 16, 18, 20, 21, 26, 30]. The key idea behind these algorithms is to mark packets proactively, that is, before congestion occurs, to limit the amount of lost traffic in the network. For instance, RED [18] and its extensions, e.g. 26] use a smoothed average of the buffer occupancy, Q, to infer impending congestion. If ....

W.-C. Feng, D. Saha, D. Kandlur, and K. Shin. A self-configuring RED gateway. In Proceedings of IEEE INFOCOM '99, volume 3, pages 1320--1328, New York, NY, March 1999. 19

.... (or equivalently sending rates) Several active queue management mechanisms have been proposed, e.g. random drop [11] early packet discard [18] early random drop [14] random early detection (RED) 8] and its variations (FRED [17] stabilized RED (SRED) 20] and balanced RED (BRED) 1] BLUE [7], REM [2] PI controller [12] and AVQ [16] among which RED has perhaps received the most attention. RED has been shown to successfully prevent global synchronization , reduce packet loss ratios, and minimize the bias against bursty sources. The major pitfalls of RED are, however, that its ....

.... the level of network load and the parameter settings, that the queue length usually oscillates significantly [3] and that it cannot achieve simultaneously high link utilization and low packet losses (due to the fact that RED directly couples the packet dropping probability with the queue length) [7], 2] 16] In the past few years, significant research efforts have been made to study improve the performance of RED (or, in general, AQM) These efforts can be roughly classified into three categories. In the first category, approaches are devised to adaptively, on line adjust RED parameters ....

[Article contains additional citation context not shown here]

W. Feng, K. Shin, D. Kandlur, and D. Saha. A Self-Configuring RED Gateway. Proceedings of IEEE INFOCOM'99, New York, USA, March 1999.

....properly. Moreover, the marking aggressiveness of the algorithm is insensitive to the number of active flows on the gateway. When the bottleneck link is shared equally between the active flows, marking one packet to send congestion feedback decreases the total throughput by the rate of 1 1 (2N) [14]. Since only the average queue length and max p parameter are used in calculation of RED s packet marking probability, RED cannot mark packets proportionally to the number of active flows. This may result in the loss of queue control. Feng proposed a modification to the RED algorithm in order to ....

....of RED s packet marking probability, RED cannot mark packets proportionally to the number of active flows. This may result in the loss of queue control. Feng proposed a modification to the RED algorithm in order to adjust its parameters dynamically according to the load on the gateway in [14]. With this modification, marking probability is increased by factor of # upon queue overflow event, and decreased by factor of # upon queue idle event. Feng verified the effectiveness of this algorithm by simulations. When large numbers of flows are active at gateways, RED causes high loss ....

Feng, W,. Kandlur, D., Shin, K., 1999, A Self-configuring RED Gateway, INFOCOM '99

.... an exponential increase in network traffic, the IETF is considering the deployment of explicit congestion notification (ECN) 11, 24, 25] along with active queue management techniques such as RED (Random Early Detection) 2, 11] While ECN is necessary for eliminating packet loss in the Internet [10], this paper shows that RED, even when used in conjunction with ECN, is ineffective in preventing packet loss. The basic idea behind RED queue management is to detect incipient congestion early and to convey congestion notification to the end hosts, allowing them to reduce their transmission ....

....to congestion notification. RED also must ensure that congestion notification is given at a rate which sufficiently suppresses the transmitting sources without underutilizing the link. Unfortunately, when a large number of TCP sources are active, the aggregate traffic generated is extremely bursty [7, 10]. Bursty traffic often defeats the active queue management techniques used by RED since queue lengths grow and shrink rapidly, well before RED can react. Figure 1 shows a simplified pictorial example of how RED functions under this congestion scenario. The congestion scenario presented in Figure ....

[Article contains additional citation context not shown here]

W. Feng, D. Kandlur, D. Saha, and K. Shin. A Self-Configuring RED Gateway. In Proc. IEEE INFOCOM, March 1999.

....max p at average queue size max th ; subsequently all packets are dropped w.p. 1. While optimum values of RED parameters as well as enhancements to the preliminary This work was supported in part by AFOSR Grant F49620 1 0472 RED algorithm [4] are still subject to ongoing research (e.g. 7] [8]) the key objective of RED can be stated as follows: packet drop (i.e. congestion notification) must be generated at a rate sufficient to prevent packet loss due to buffer overflow (otherwise, RED will act as a Tail Drop) while sustaining high link utilization. In view of the Internet ....

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," in Proceedings of INFOCOM'99, 1999.

....queue size. The weighting factor of the EWMA method is denoted as . As a reminder the RED function is depicted in figure 1. Since the publishing of [5] significant research has been devoted to the analysis of the dynamics of RED in interaction with TCP congestion control. Several papers, [3], 4] 5] 10] 12] introduce models aimed at finding an appropriate set of the parameters. The goal of a proper RED adjustment is to maintain the average queue size between and at low oscillations and in turn help avoid forced drops (i.e. either the instantaneous queue size ....

W. Feng, D. Kandlur, D. Saha, and K. Shin. A Self-Configuring RED Gateway. Infocom, March 1999.

....we turn it on for all simulations in this paper. Moreover, in this paper, we only consider RED without support for ECN. 5.2. RED Configurations RED has four control parameters: min th , max th , max p , and w q . How to properly configure these parameters has been the subject of many studies [8, 13, 11, 18]. The focus of our work is on understanding the impact of RED on TCP performance. Therefore, instead of proposing any new recommendations on configuring RED parameters, we closely follow the guidelines by Floyd [18] More specifically, we always use the recommended values of max th = 3 min th , ....

.... dropping policy, ranging from analytical modeling [5, 31] to simulations and experimental evaluation [45, 10, 8, 48] from recommendations for parameters and architecture configuration [18, 19, 13, 22] to proposals of alternatives, which include BLUE [12] SRED (Stabilized RED) 34] Adaptive RED [11], FRED (Fair Random Early Drop) 28] and BRED (Balanced RED) 4] Our simulation study of RED performance complements the existing body of research by comparing the aggregate throughput, goodput, loss rate, and roundtrip bias of TCP connections under RED gateways with those under drop tail ....

W. Feng, D. Kandlur, D. Saha, and K. Shin. A SelfConfiguring RED Gateway. Proc. IEEE INFOCOM'99, New York City, USA, March 1999.

.... caused by an exponential increase in network traffic, the IETF is considering the deployment of explicit congestion notification (ECN) 12] along with active queue management techniques such as RED (Random Early Detection) 1] While ECN is necessary for eliminating packet loss in the Internet [6], this paper shows that RED, even when used in conjunction with ECN, is ineffective in preventing packet loss. The basic idea behind RED queue management is to detect incipient congestion early and to convey congestion notification to the end hosts, allowing them to reduce their transmission ....

....to congestion notification. RED also must ensure that congestion notification is given at a rate which sufficiently suppresses the transmitting sources without underutilizing the link. Unfortunately, when a large number of TCP sources are active, the aggregate traffic generated is extremely bursty [4, 6]. Bursty traffic often defeats the active queue management techniques used by RED since queue lengths grow and shrink rapidly, well before RED can react. Figure 1 shows a simplified pictorial example of how RED functions under this congestion scenario. The congestion scenario presented in Figure ....

[Article contains additional citation context not shown here]

W. Feng, D. Kandlur, D. Saha, and K. Shin. A SelfConfiguring RED Gateway. In Proc. IEEE INFOCOM, March 1999.

.... an exponential increase in network traffic, the IETF is considering the deployment of explicit congestion notification (ECN) 12, 30, 31] along with active queue management techniques such as RED (Random Early Detection) 3, 14] While ECN is necessary for eliminating packet loss in the Internet [9], we show that RED, even when used in conjunction with ECN, is ineffective in preventing packet loss. The basic idea behind RED queue management is to detect incipient congestion early and to convey congestion notification to the end hosts, allowing them to reduce their transmission rates before ....

....to congestion notification. RED must also ensure that congestion notification is given at a rate which sufficiently suppresses the transmitting sources without underutilizing the link. Unfortunately, when a large number of TCP sources are active, the aggregate traffic generated is extremely bursty [8, 9]. Bursty traffic often defeats the active queue management techniques used by RED since queue lengths grow and shrink rapidly, well before RED can react. Figure 1(a) shows a simplified pictorial example of how RED functions under this congestion scenario. The congestion scenario presented in ....

[Article contains additional citation context not shown here]

W. Feng, D. Kandlur, D. Saha, and K. G. Shin. A Self-Configuring RED Gateway. In Proc. IEEE INFOCOM, pages 1320--1328, March 1999.

....probability of the nonresponsive flow is 1.0, it is identified as being nonresponsive and rate limited. Note that just as BLUE s marking probability can be used in SFB to provide protection against nonresponsive flows, it is also possible to apply Adaptive RED s max p parameter to do the same [5]. In this case, a per bin max p value is kept and updated according to the behavior of flows which map into the bin. As with RED, however, there are two problems which make this approach ineffective. The first is the fact that a large amount of buffer space is required in order to get RED to ....

W. Feng, D. Kandlur, D. Saha, and K. Shin. A SelfConfiguring RED Gateway. In Proc. IEEE INFOCOM, March 1999.

....avoiding bias against bursty traffic. Furthermore, the bursty traffic in the refined RED also benefits from the avoidance of buffer overflows. It has less timeouts and higher throughput than the bursty traffic in the original RED. VI. RELATED WORK A self configuring RED gateway was proposed in [3], which adaptively changes the maximum drop probability a depending on the variations in average queue size. However, only , which indicates the aggressiveness of RED, is adaptively changed; which indicates the responsiveness of RED, is still kept constant. Also, the adaptive change ....

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin, "A Self-Configuring RED Gateway", Proceedings of IEEE INFOCOM'99, New York, NY, pp. 1320-1328, March 1999.

....of competing connections (N ) The lack of knowledge of N can cause congestion notification to be either too aggressive or not aggressive enough. Generally, when a bottleneck link is shared between N users, congestion notification to one connection reduces the o#ered load by a factor of (1 2N ) [2]. This can be a paltry drop of 0.05 when 100 TCP connections share the link, while resulting in a severe 25 drop when there are only two connections. BLUE [1] uses packet loss and link idle events in addition to queue size to avoid congestion. BLUE varies the marking probability, Pm , depending ....

W. Feng, D. Kandlur, D. Saha, K. Shin. A SelfConfiguring RED Gateway. In Proceedings of IEEE INFOCOM, March 1999.

No context found.

W. Feng, D. Kandlur, D. Saha, K. Shin, "A Self-Configuring RED Gateway", in Proc. of IEEE Infocom, New York, NY, 1999, pp.1320-1328.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," in Proc. IEEE INFOCOM'99, March 1999.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin. A self-configuring RED gateway. In Proc. IEEE INFOCOM'99, March 1999. 26

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proc. of INFOCOM '99, pages 1320--1328, March 1999.

No context found.

Wu-chang Feng, Dilip D. Kandlur, Debanjan. Saha, and Kang. G. Shin. A self-configuring RED gateway. In Proceedings of 18th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM), pages 1320--1328, March 1999.

No context found.

W. Feng, D. Kandlur, D. Saha and K. Shin, "A self-configuring RED gateway," Proceedings of INFOCOM '99, pp. 1320-1328, Mar. 1999.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin. A self--configuring RED gateway. In Proceedings of INFOCOM'99, March 1999. http://www.eecs.umich.edu/~wuchang/work/infocom99. ps.Z.

No context found.

Wu-Chang Feng, Dilip D. Kandlur, Debanjan Saha, and Kang G. Shin. A Self-Configuring RED Gateway. In Proceedings of IEEE INFOCOM 1999.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," in Proceedings of INFOCOM 99, vol. 3, 1999, pp. 1320--1328.

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin, "A self-configuring RED gateway," in Proc. IEEE INFOCOM, 1999.

No context found.

W. Feng, D. Kandlur, D. Saha and K. Shin, "A Self-Configuring RED Gateway", Proceedings of INFOCOM'99: Conference on Computer Communications 1999, p.1320-8 vol.3.

No context found.

Wu-chang Feng and Dilip D.Kandlur and Debanjan Saha and Kang G. Shin, "A self-configuring RED gateway," in Proceedings of IEEE INFOCOM'99, March 1999.

No context found.

Feng, W,. Kandlur, D., Shin, K., 1999, A Self-configuring RED Gateway, INFOCOM '99

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proceedings of the 17th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM '99), pages 1320--1328, March 1999.

No context found.

W.-C. Feng, D. Saha, D. Kandlur, and K. Shin. A self-configuring RED gateway. In Proceedings of IEEE INFOCOM '99, volume 3, pages 1320--1328, New York, NY, March 1999.

No context found.

Feng W, Kandlur D, Saha D, Shin K (1999) A self-configuring RED gateway. IEEE INFOCOM'99, New York

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A SelfConfiguring RED Gateway", IEEE Infocom, March 1999.

No context found.

W. chang Feng, D. D. Kandlur, D. Saha, and K. G. Shin, "A self-configuring RED gateway," in Proc. INFOCOM'99, vol. 3. IEEE, Mar. 1999, pp. 1320--1328.

No context found.

Wu-chang Feng and Dilip D.Kandlur and Debanjan Saha and Kang G. Shin, "A self-configuring RED gateway," in Proceedings of IEEE INFOCOM'99, March 1999.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," IEEE INFOCOM'99, New York, Mar. 1999.

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proc. of INFOCOM '99, pages 1320--1328, March 1999.

No context found.

W. chang Feng, D. D. Kandlur, D. Saha, and K. G. Shin, "A self-configuring RED gateway," in Proceedings of IEEE INFOCOM '99, March 1999.

No context found.

W. Feng, D. D. Kandlur, D. Saha, K. G. Shin, A self-configuring RED gateway, in: Proceedings of INFOCOM 99, Vol. 3, 1999, pp. 1320--1328.

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proc. of INFOCOM '99, pages 1320--1328, March 1999.

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," in Proceedings of INFOCOM 99, vol. 3, 1999, pp. 1320--1328.

No context found.

Wu chang Feng, Dilip D. Kandlur, Debanjan Saha, and Kang G. Shi. A self-configuring RED gateway. Proceedings of IEEE Infocom 1999.

No context found.

W. Feng, D. Kandlur, D. Saha and K. Shin, A self-configuring RED gateway, Proceedings of INFOCOM '99, Eighteenth Annual Joint Conference of the IEEE Computer and Communications Societies, Vol. 3 (1999), pp 1320-1328.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A self-configuring RED gateway," in Proc. o3 INFOCOM, pp. 1320-1328, March 1999.

No context found.

W. Feng, D. D. Kandlur, D. Saha, and K. G. Shin. A self-configuring RED gateway. In Proc. of INFOCOM '99, pages 1320--1328, March 1999.

No context found.

W. Feng, D. Kandlur, D. Saha and K. Shin, "A Self-Configuring RED Gateway", Proceedings of INFOCOM'99: Conference on Computer Communications 1999, p. 1320-8 vol.3.

No context found.

W. Feng, D. Kandlur, D. Saha, and K. Shin, "A Self-Configuring RED Gateway," in Proceedings of the IEEE INFOCOM, Mar. 1999.

First 50 documents  Next 50

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