A Performance Study of Loss Detection/Recovery in Real-world TCP Implementations

Rewaskar, Sushant; Kaur, Jasleen; Smith, F. Donelson

doi:10.1109/icnp.2007.4375856

Cited by 29 publications

(21 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…While RFC 2988 [37] recommends setting the minRTO to 1 sec, the current practice is setting minRTO to 0.2 sec [40]. Hence, we repeat the experiment with minRTO 0.2 sec and initRTO 3.0 sec, and we obtain results consistent to those shown in Figure 2.…”

Section: Sub-exponential Backoffsupporting

confidence: 76%

See 1 more Smart Citation

Removing exponential backoff from TCP

Mondal

Kuzmanovic

2008

SIGCOMM Comput. Commun. Rev.

View full text Add to dashboard Cite

The well-accepted wisdom is that TCP's exponential backoff mechanism, introduced by Jacobson 20 years ago, is essential for preserving the stability of the Internet. In this paper, we show that removing exponential backoff from TCP altogether can be done without inducing any stability sideeffects. We introduce the implicit packet conservation principle and show that as long as the endpoints uphold this principle, they can only improve their end-to-end performance relative to the exponential backoff case.By conducting large-scale simulations, modeling, and network experiments in Emulab and the Internet using a kernellevel FreeBSD TCP implementation, realistic traffic distributions, and complex network topologies, we demonstrate that TCP's binary exponential backoff mechanism can be safely removed. Moreover, we show that insuitability of TCP's exponential backoff is fundamental, i.e., independent from the currently-dominant Internet traffic properties or bottleneck capacities. Surprisingly, our results indicate that a path to incrementally deploying the change does exist.

show abstract

Section: Sub-exponential Backoffsupporting

confidence: 76%

“…The authors acknowledge that at some future point, research may show that a smaller RTO is acceptable or superior. To the best of our knowledge, the current practice in most operating systems is setting the minimum RTO parameter to 200 msec [40].…”

Section: Retransmission Timeoutsmentioning

confidence: 99%

Removing exponential backoff from TCP

Mondal

Kuzmanovic

2008

SIGCOMM Comput. Commun. Rev.

View full text Add to dashboard Cite

show abstract

“…There is much prior work with regards to reconstructing TCP flow from passive measurements and using this information to understand the end-toend properties of traffic [29,30,31,21]. However, the MAWI traces impose two constraints which require careful consideration, and ultimately imply the use of a custom TCP tracer.…”

Section: Tracing Tcp Metricsmentioning

confidence: 99%

On rate limitation mechanisms for TCP throughput: A longitudinal analysis

et al. 2017

View full text Add to dashboard Cite

TCP remains the dominant transport protocol for Internet traffic. It is usually considered to have its sending rate covered by a sliding window congestion control mechanism. However, in addition to this normal congestion control, a number of other mechanisms limit TCP throughput. This paper analyzes the extent to which network, host and application settings define flow throughput over time and across autonomous systems. Our study draws on data from a longitudinal study spanning five years of passive traces collected from a single transit link. Mechanisms for this include limiting by application, interference with the TCP window control mechanism and artificial limitations on maximum window sizes by the operating system. This paper uses a large data set to assess the impact of each mechanism. We conclude that more than half of all heavy-hitter inbound traffic remains throttled by constraints beyond network capacity. For this data set, TCP congestion control is no longer the dominant mechanism that moderates throughput.

show abstract

“…We employed TCP Reno [22] with 10 ms granularity timers, as on most modern systems [20]; the coarse TCP timer granularity of older systems would make TCP react unrealistically slow to losses. In File Transfer, a file was transmitted from the server to the client.…”

Section: Wired Servermentioning

confidence: 99%

Link Layer Adaptation for Shared Wireless Links

Xylomenos¹,

Makidis²

2008

Mobile Netw Appl

View full text Add to dashboard Cite

Abstract-While traditional link layer protocols assume that they fully control the underlying link, in contemporary wireless networks the link may be dynamically shared by sessions belonging to different users and/or applications. To assess the impact of link sharing, we measure the File Transfer and Web Browsing throughput achieved over a Selective Repeat (SR) protocol, with or without contention from Media Distribution. Our results indicate that the optimal protocol settings strongly depend on the level of contention for the link. We therefore present two link layer protocols that adapt to the available bandwidth, our Adaptive Selective Repeat (ASR) protocol which dynamically modifies its retransmission timeouts, and the Radio Link Control (RLC) protocol specified for use by UMTS networks which does not employ retransmission timers. We first repeat our performance measurements to determine the optimal settings for each protocol, and then compare the fine tuned versions of all protocols with respect to their File Transfer and Web Browsing throughput, as well as to the delay induced to the contending Media Distribution packets. Our results indicate that while both RLC and ASR are more stable than SR, the complex RLC does not match the performance of our simpler ASR. Index Termsink Layer Protocols, Adaptive Selective Repeat, Radio Link Controlink Layer Protocols, Adaptive Selective Repeat, Radio Link ControlL

show abstract

A Performance Study of Loss Detection/Recovery in Real-world TCP Implementations

Cited by 29 publications

References 21 publications

Removing exponential backoff from TCP

Removing exponential backoff from TCP

On rate limitation mechanisms for TCP throughput: A longitudinal analysis

Link Layer Adaptation for Shared Wireless Links

Contact Info

Product

Resources

About