14 research outputs found
Another Examination of the Use-it-or-Lose-it Function on TCP Traffic
No full text: We review the new proposed End-System Method 1 for Use-it-or-lose-it behavior proposed in Annex F of the Traffic Management Specification Version 4.0, R10, for bi-directional TCP traffic. We have studied this in an environment of explicit rate switches. This contribution is a follow-up to the work discussed in ATM Forum/951598. NOTICE: This document has been prepared to assist the ATM Forum's Traffic Management Working Group. This document is offered by the University of California and AT&T to the ATM Forum as a basis for discussion and is not a binding proposal on either the University of California system or AT&T. Any requirements stated herein are subject to change in form and content after more study. UC Santa Cruz and AT&T reserve the right to add to, amend, or withdraw the statements contained herein. Another Examination of the Use-it-or-Lose-it Function on TCP Traffic Lampros Kalampoukas Anujan Varma University of California at Santa Cruz K. K. Ramakrishnan Kerry Fendick AT..
Performance of TCP over Multi-Hop ATM Networks: A Comparative Study of ATM-Layer Congestion Control Schemes
No full textthis paper is to study the dynamic behavior of TCP over ATM networks and internetworks consisting of multiple hops of ATM switches, in an effort to identify the influence of various ATM-layer congestion control approaches on TCP performance. Of particular interest is to investigate the interaction between connections with different round-trip delays to determine how the fairness characteristics vary for the different schemes. To achieve these objectives, we simulated a network configuration with 8 nodes, 4 switches, and a total of 10 TCP sessions sharing network resources. In addition to the baseline case with no congestion control, we considered two distinct congestion control approaches at the ATM layer --- the ATM Early Packet Discard (ATM-EPD) proposed by Romanow and Floyd [15] and link-level flow control proposed by Kung and Chapman [10] and Varghese, et al. [17]. To compare with the performance of TCP over a datagram network, we also simulated a network configuration in which the ATM switches were replaced by IP gateways. The rest of this paper is organized as follows: In Section 2, we outline the ATM-layer congestion control approaches simulated. In Section 3 we describe the network model and tools used in the simulations. We discuss the simulation results in Section 4 and compare the congestion-control approaches in terms of their effective throughput, fairness in bandwidth usage, number of retransmissions, and delay characteristics. In Section 5 we discuss the problem of setting the threshold in the ATM Early Packet Discard scheme. We conclude the paper in Section 6 with a discussion of the lessons learned and directions for future research. Appendix A provides details of computing the buffer sizes for the link-level flow control scheme we simulated. 2 Congesti..
Two-Way TCP Traffic over ATM: Effects and Analysis
No full textthis paper, our interest is in analyzing the dynamics of TCP connections in an Asynchronous Transfer Mode (ATM) network 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 between the same pair of end nodes over a network path. The TCP segments transmitted by the connections in one direction share the same physical path with the acknowledgements (acks) of connections in the opposite direction. These packets and acknowledgements may share a common buffer in the end systems as well as network switches/routers. This sharing has been shown to result in an effect called ack compression, where acks of a connection arrive at the source bunched together [WRM91, ZSC91]. The result of ackcompression is a marked unfairness in the throughput received with competing connections, and reduced overall throughput compared to what could be expected without this effect [WRM91]. Ack compression may occur either at the end system or in a switch/router. In either case, the smooth flow of acknowledgements to the source is disturbed, potentially resulting in reduction of throughput for the TCP connections involved. The effect of ack compression and the resulting dynamics of transport protocols under two-way traffic have been studied previously by Zhang, et al. [ZSC91], and by Wilder, et al. [WRM91]. Zhang, et al. [ZSC91] studied TCP dynamics under two-way traffic in a datagram network by simulation, and observed that the queues in the routers exhibit periodic behavior. Wilder, et al. [WRM91] observed a similar effect in OSI-based networks under two-way traffic causing unfairness and an overall reduction in throughput. While these studies provide a qualitative tr..
Two-Way TCP Traffic over Rate Controlled Channels: Effects and Analysis
No full textWe examine the performance of bidirectional TCP/IP connections over a network which uses ratebased flow and congestion control mechanisms. An example of such a network is an Asynchronous Transfer Mode (ATM) network using the Available Bit Rate (ABR) service. The sharing of a common buffer by TCP packets and acknowledgements has been known to result in an effect called ack compression, where acks of a connection arrive at the source bunched together, resulting in unfairness and degraded throughput. It has been the expectation that maintaining a smooth flow of data using rate-based flow control would mitigate, if not eliminate, the various forms of burstiness experienced with the TCP window flow control. However, we show that the problem of TCP ack compression re-appears even while operating over a rate-controlled channel, although the queues are primarily at the end systems now. By analyzing the periodic bursty behavior of the source IP queue, we are able to predict the peak values for..
Explicit Window Adaptation: A Method to Enhance TCP Performance
No full textWe study the performance of TCP in an internetwork consisting of both rate-controlled and non-ratecontrolled segments. A common example of such an environment occurs when the end systems are part of IP datagram networks interconnected by a rate-controlled segment, such as an ATM network using the ABR service. In the absence of congestive losses in either segment, TCP keeps increasing its window to its maximum size. Mismatch between the TCP window and the bandwidth-delay product of the network will result in accumulation of large queues and possibly buffer overflows in the devices at the edges of the rate-controlled segment, causing degraded throughput and unfairness. We develop an explicit feedback scheme, called Explicit Window Adaptation based on modifying the receiver's advertised window in TCP acknowledgments returning to the source. The window size indicated to TCP is a function of the free buffer in the edge device. Results from simulations with a wide range of traffic scenarios ..
Improving TCP Throughput over Two-Way Asymmetric Links: Analysis and Solutions
No full textThe sharing of a common buffer by TCP data segments and acknowledgments in a network or internet has been known to produce the effect of ack compression, often causing dramatic reductions in throughput. We study several schemes for improving the performance of two-way TCP traffic over asymmetric links where the bandwidths in the two directions may differ substantially, possibly by many orders of magnitude. These approaches reduce the effect of ack compression by carefully controlling the flow of data packets and acknowledgments. We first examine a scheme where acknowledgments are transmitted at a higher priority than data. By analysis and simulation, we show that prioritizing acks can lead to starvation of the low-bandwidth connection. Next, we introduce and analyze a connection-level backpressure mechanism designed to limit the maximum amount of data buffered in the outgoing IP queue of the source of the low-bandwidth connection. We show that this approach, while minimizing the queuei..
Dynamics of an Explicit Rate Allocation Algorithm for Available Bit-Rate (ABR) Service in ATM Networks
No full textthis paper we study the performance of an explicit rate allocation algorithm for ATM Networks using the available bit-rate (ABR) class of service. We examine the behavior of ABR traffic with simple cell sources, and demonstrate that the allocation algorithm is fair and maintains network efficiency in a variety of network configurations. We also study the behavior of TCP/IP sources using ABR service in a network of switches employing the rate allocation algorithm; the results show substantial improvements in fairness and efficiency in comparison with the performance of TCP on an underlying datagram-based network. Even in a dynamic network configuration with frequent opening and closing of ABR connections, TCP connections were able to make sustained progress with a switch buffer size as small as 1 Kbyte, providing an average link utilization of approximately 60% of the attainable maximum. We also study the performance of ABR traffic in the presence of higher-priority variable bit-rate (VBR) video traffic and show that the overall system achieves high utilization with modest queue sizes in the switches, and the ABR flows adapt to the available rate in a fairly short interval. Scalability of the ABR service is an important criterion. We demonstrate the scalability of the rate allocation algorithm with respect to the number of connections by increasing the number of active connections with very long round-trip times by a factor of 13; this caused only a three times increase in the switch queue size, while still maintaining maximal link utilization. Keywords: Explicit rate allocation, congestion control, TCP over ATM This research is supported by the Advanced Research Projects Agency (ARPA) under Contract No. F19628-93C -0175 and by the NSF Young Investigator Award No. MIP-92..
