L. Zhang and D. Clark. Oscillating behavior of network traffic: a case study simulation. Internetworking:research and experience, 1(2):101--112, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....developments in Active Queue Management (AQM) techniques have introduced additional complexity in transport protocols. The development of AQM was prompted by the observation that with simple Tail Drop gateways, TCP congestion control leads to undesirable behavior, i.e. global synchronization [7]. When several TCP flows compete for bandwidth in a Tail Drop gateway, it has been observed experimentally that packets from many flows are usually discarded simultaneously, resulting in a poor utilization of the network. AQM algorithms such as Random Early Detection (RED) 8] and Explicit ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetworking: Research and Experience,vol.1, no. 2, pp. 101--112, 1990.

....arrive to a full buffer. When multiple TCP sessions sharing a common gateway arrive at a full buffer, the impact is (typically) simultaneous packet loss and reduction of transmission rate in all flows, resulting in an oscillatory behavior termed global synchronization reported experimentally by [1], 2] This leads to link capacity under utilization and exacerbates the inherent bias of TCP congestion control algorithm against higher delay flows. Accordingly, packet drop policies alternate to Tail Drop have been sought a promising class is based on randomized packet drop [3] 4] or ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: a case study simulation," Internetworking: Research and Experience, vol. 1, no. 2, pp. 101--12, 1990.

....to its round trip delay. Proof: We show below that, for a Tail Drop queue, the steady state window size for any flow is inverse proportional to the round trip time. In Tail Drop queues, buffer overflow causes (typically) at least one packet loss from each flow passing through the queue ([35], 36] Using the same notations as Sections II and III, 60) Taking expectation of both sides of (60) and denoting W j = 2 (61) proving the claim. ....

L. Zhang and D. Clark. Oscillating behavior of network traffic: a case study simulation. Internetworking: Research and Experience, 1(2):101--12, 1990.

....that with simple Tail Drop gateways, TCP congestion control leads to undesirable behavior, i.e. global synchronization. When several TCP flows compete for bandwidth in a TailDrop gateway, it has been observed experimentally that packets from many flows are usually discarded simultaneously [23], resulting in a poor utilization of the network. AQM algorithms such as Random Early Detection (RED) 10] and Explicit Congestion Notification (ECN) 7] have been proposed to help alleviate this problem by randomly dropping marking packets depending on queue size. This allows each TCP flow to ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, 1990.

....fact that with simple Tail Drop gateways, TCP congestion control leads to undesirable behavior, i.e. global synchronization. When several TCP flows compete for bandwidth in a Tail Drop gateway, it has been observed experimentally that packets from many flows are usually discarded simultaneously [9], resulting in a poor utilization of the network. AQM algorithms such as Random Early Detection (RED) 10] and Explicit Congestion Notification (ECN) 11] have been proposed to help alleviate this problem by randomly dropping marking packets depending on queue size, thereby avoiding heavy ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, 1990.

....known that with taildrop gateways this congestion control also leads to undesirable behavior, i.e. global synchronization. When several TCP flows compete for bandwidth in a tail drop gateway, it has been observed experimentally that packets from many flows are usually discarded simultaneously [1], resulting in a poor utilization of the network. Active queue management algorithms such as Random Early Detection (RED) 2] were introduced to help alleviate this problem by randomly dropping packets depending on the queue size, thereby avoiding heavy congestion and prevent global ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, 1990.

....with tail drop gateways this congestion control mechanism also leads to undesirable behavior, i.e. global synchronization. When several TCP flows compete for bandwidth in a tail drop gateway, it has been observed experimentally that packets from many flows are usually discarded simultaneously [1], resulting in a poor utilization of the network. Active queue management algorithms such as Random Early Detection (RED) 2] were introduced to help alleviate this problem by randomly dropping packets depending on the queue size, thereby avoiding heavy congestion and preventing global ....

L. Zhang and D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, 1990.

....in the network. Usually, losses occur if the buffer at a network gateway has reached its capacity and every incoming packet is dropped (e.g. drop tail gateways) In TCP, there are no means to discover congestion before a packet is lost. This leads to the problem of global synchronization [9]. When a gateway starts to drop packets due to a buffer overflow, many flows see a dropped packet at roughly the same time. Consequently, they all reduce their congestion window at the same time, which leads to a low utilization after a busy cycle and causes a greater variance in the queuing ....

L. Zhang and D. Clark. Oscillating Behavior of Network Traffic: A Case Study Simulation. Internetworking: Research and Experience, 1(2):101--112, December 1990.

....arrive at the gateway [7] Another goal in deciding which connections to notify of congestion is to avoid the global synchronization that results from notifying all connections to reduce their windows at the same time. Global synchronization has been studied in networks with Drop Tail gateways [37], and results in loss of throughput in the network. Synchronization as a general network phenomena has been explored in [8] In order to avoid problems such as biases against bursty traffic and global synchronization, congestion avoidance gateways can use distinct algorithms for congestion ....

Zhang, L., and Clark, D., "Oscillating Behavior of Network Traffic: A Case Study Simulation", Internetworking: Research and Experience, Vol. 1, 1990, pp. 101-112.

....when the sources are Web users. In the context of single bottleneck link networks and infinite source models, TCP IP traffic synchronization effects have been well studied in the past and have led to many influential papers on understanding the dynamics of TCP type congestion control algorithms, [14, 18, 22, 23]. We contribute in this paper to the existing body of knowledge by demonstrating how accounting for realistic workload heterogeneity (via appropriate TCP connection level variability) can yield new insights into actual TCP behavior. In fact, using the terminology originally due to V. Jacobson, we ....

.... we also experiment with a link delay of 0:14 seconds) We chose this admittedly unrealistic network configuration because we wanted to illustrate our findings in a setting that has been used (with different parameter values, though) in a number of previous studies of the dynamics of TCP (e.g. [18, 22, 23]) 2.2 Two different workload models In contrast to our homogeneous network configuration, our workload models for the clients are heterogeneous in nature and allow for a spectrum of variability at the user level. On the one side of the spectrum, we deal with the case of no user level ....

L. Zhang and D. Clark. Oscillating Behavior of Network Traffic: A Case Study Simulation. Internetworking: Research and Experience, 1(2):101--112, 1990.

....rate controlled widearea network. Even when the latter is small compared to the former, Explicit Window Adaptation resulted in close to ideal throughput and fairness. The performance of TCP when operating exclusively over IP datagram networks has been the subject of extensive research in the past [6, 7, 8, 9, 10, 11, 12, 13, 14, 5]. Similarly, TCP performance over ATM networks using the Available Bit Rate (ABR) service has also been studied [15, 16, 17] These studies provide valuable insights into TCP dynamics and how TCP congestion control mechanisms interact with ABR congestion control algorithms. Several modifications ....

L. Zhang and D. D. Clark, "Oscillating behavior of network traffic: A case study simulation," Intenetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, December 1990.

.... and fairness and lower response time for delay sensitive applications (e.g. telnet) A detailed description of TCP, the algorithms it employs, and its implementation can be found in [91, 92] Several studies report on the efficiency of the mechanisms employed by TCP and its performance [9, 27, 30, 40, 98, 100]. An analytical approach to TCP performance can be found in [66] Studies on the performance of TCP over ATM networks can be found in [53, 87] Finally, performance issues of TCP over ATM for a specific TCP implementation are discussed in [21] and [78] The assumption that a packet loss is a ....

L. Zhang and D. D. Clark, "Oscillating behavior of network traffic: A case study simulation," Intenetworking: Research and Experience, vol. 1, no. 2, 101--112, December 1990. 247

....(e.g. SLIP connections) to gigabit sec connections. TCP has been designed to operate in networks with no congestion control mechanisms below the transport layer [10] It has built in mechanisms for protection against congestion. Its behavior in datagram network has been studied extensively [11, 12, 13]. However, there are several factors that justify the use of a congestion control scheme at the ATM layer, complementing in a way the mechanisms built within TCP or any other high level protocol. Without ATM layer congestion control, the performance of TCP over ATM can be considerably worse as ....

L. Zhang and D. D. Clark, "Oscillating Behavior of Network Traffic: A Case Study Simulation," Intenetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, December 1990.

....largely to the explosive growth of the TCP IP Internet in recent years. An important component of TCP is the collection of algorithms used to perform congestion control and recovery [1, 2] These algorithms give rise to a variety of interesting dynamics, some of which have been studied extensively [3, 4, 5, 6]. In this paper, our interest is in analyzing the dynamics of TCP connections over asymmetric access links in the presence of two way traffic. We define two way or bidirectional traffic as the traffic pattern resulting from two or more TCP connections transferring data in opposite directions ....

L. Zhang and D. D. Clark, "Oscillating behavior of network traffic: A case study simulation," Intenetworking: Research and Experience, vol. 1, no. 2, pp. 101--112, December 1990.

....conversation lifetimes etc. Cabrera et al. [4] have studied performance achieved by user processes in Ethernet based environments. Mills [8] has studied roundtrip transit delays over Internet as a function of packet length, and the effectiveness of TCP retransmission timeout algorithm. Zhang et al. [14] have used simulation to study network dynamics. In order to better understand the characteristics of end to end paths in a network we conducted a series of measurements on the Internet. In this report, we present the results of some of these measurements. The results presented in this report ....

Lixia Zhang and David D. Clark. Oscillating behavior of network traffic: A case study simulation. J. of Internetworking: Research & Experience,

No context found.

L. Zhang and D. Clark. Oscillating behavior of network traffic: a case study simulation. Internetworking:research and experience, 1(2):101--112, 1990.

No context found.

L. Zhang and D. Clark. Oscillating behavior of network traffic: a case study simulation. Internetworking:research and experience, 1(2):101--112, 1990.

No context found.

L. Zhang and D. Clark. Oscillating behavior of network traffic: a case study simulation. Internetworking:research and experience, 1(2):101--112, 1990.

No context found.

L.Zhang, D.Clark, "Oscillating Behavior of Network Traffic: a Case Study Simulation", Internetworking: Research and Experience, vol. 1, n. 2, pp. 101-112, 1990.

No context found.

L. Zhang and D. D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetwoking: Research and Experience, vol. 1, pp. 101-- 112, 1990.

No context found.

L. Zhang, D. D. Clark, and Scott, "Oscillating behavior of network traffic: A case study simulation," Intetnetworking: Research and Experience, vol. 1, pp. 101--112, 1990.

No context found.

L. Zhang and D Clark. Oscillating Behavior of Network Traffic: A Case Study Simulation. Internetworking: Research and Experience, 1(2):101--112, December 1990. (p 159)

No context found.

L.Zhang, D.Clark, "Oscillating Behavior of Network Traffic: a Case Study Simulation", Internetworking: Research and Experience, vol. 1, n. 2, pp. 101-112, 1990.

No context found.

Lixia Zhang and David D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetwoking: Research and Experience, vol. 1, pp. 101--112, 1990.

No context found.

Lixia Zhang and David D. Clark, "Oscillating behavior of network traffic: A case study simulation," Internetwoking: Research and Experience, vol. 1, pp. 101--112, 1990.

First 50 documents

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