*BSD News Article 48406


Return to BSD News archive

Xref: sserve comp.os.linux.networking:11441 comp.os.ms-windows.networking.tcp-ip:13667 comp.unix.bsd.freebsd.misc:4387
Path: sserve!euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news.sprintlink.net!howland.reston.ans.net!nntp.crl.com!pacbell.com!nntp-hub2.barrnet.net!parc!fenner
From: fenner@parc.xerox.com (Bill Fenner)
Newsgroups: comp.os.linux.networking,comp.os.ms-windows.networking.tcp-ip,comp.unix.bsd.freebsd.misc
Subject: Re: Internet service providing-which OS?
Date: 4 Aug 1995 20:55:41 GMT
Organization: Xerox Palo Alto Research Center
Lines: 13
Distribution: best
Message-ID: <3vu1gd$7gg@news.parc.xerox.com>
References: <3ue5qa$ain@panix.com> <3urba1$cgl@blob.best.net> <3us870$15o@agate.berkeley.edu> <B.A.MCCAULEY.95Aug3145624@wcl-l.bham.ac.uk>
NNTP-Posting-Host: crevenia.parc.xerox.com

In article <B.A.MCCAULEY.95Aug3145624@wcl-l.bham.ac.uk>,
 <B.A.McCauley@bham.ac.uk> wrote:
>Actually I think Ethernet = 2 T1s is a better approximation. I seem to
>recall that with a random traffic profile CSMA/CD (ethernet) actually
>saturates with an average traffic at 1/e (about 37%) of the burst
>transfer rate. Beyond that, the collision retry queues just keep on
>growing.

This is a widely spread myth.  See, for example, the WRL tech report 88.4,
"Measured Capacity of an Ethernet: Myths and Reality", by Boggs, Mogul and
Kent.  <http://www.research.digital.com/wrl/techreports/abstracts/88.4.html>

  Bill