ACM/IEEE SC 2002 Conference (SC'02) 2002
DOI: 10.1109/sc.2002.10023
|View full text |Cite
|
Sign up to set email alerts
|

A TCP Tuning Daemon

Abstract: Many high performance distributed applications require high network throughput but are able to achieve only a small fraction of the available bandwidth. A common cause of this problem is improperly tuned network settings. Tuning techniques, such as setting the correct TCP buffers and using parallel streams, are well known in the networking community, but outside the networking community they are infrequently applied. In this paper, we describe a tuning daemon that uses TCP instrumentation data from the Unix ke… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
50
0

Year Published

2003
2003
2010
2010

Publication Types

Select...
4
2
2

Relationship

0
8

Authors

Journals

citations
Cited by 42 publications
(50 citation statements)
references
References 28 publications
0
50
0
Order By: Relevance
“…The optimal buffer size is related to bandwidth-delay product which is the product of the data link capacity (bandwidth of the bottleneck link) with end-to-end delay (the round trip time of the connection) [5,6,7]. There have been many studies that investigate diagnosis techniques to measure bandwidth and round-trip time of a given connection [4]. The bandwidth-delay product can vary in a shared wide-area network environment.…”
Section: Related Work and Discussionmentioning
confidence: 99%
“…The optimal buffer size is related to bandwidth-delay product which is the product of the data link capacity (bandwidth of the bottleneck link) with end-to-end delay (the round trip time of the connection) [5,6,7]. There have been many studies that investigate diagnosis techniques to measure bandwidth and round-trip time of a given connection [4]. The bandwidth-delay product can vary in a shared wide-area network environment.…”
Section: Related Work and Discussionmentioning
confidence: 99%
“…Other media support a larger MTU: FDDI has a 4392-byte MTU, GigE jumbo-frames have a 9000-byte MTU, IP-over-ATM uses a 9180-byte MTU, and HiPPI uses a 65535-byte MTU. However, these are being phased out in favor of higher speed Ethernet [7,11].…”
Section: Frame Sizementioning
confidence: 99%
“…As network throughput capacity has increased in recent years, operating systems have gradually changed the default buffer size from common values of 8 kilobytes to as much as 64 KBytes. However, this is still far too small for today's high speed networks [11], and prevents TCP from using all the bandwidth available.…”
Section: Tcp Buffersmentioning
confidence: 99%
“…The projects developed so far, like WAD (Work Around Daemon) [1] or ENABLE [2] don't meet the imposed requirements, as for WAD a modified kernel must be used, and for ENABLE the applications must be rewritten.…”
Section: Introductionmentioning
confidence: 99%