Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ENG.SUN.COM!nowicki
From: nowicki@ENG.SUN.COM (Bill Nowicki)
Newsgroups: comp.protocols.tcp-ip
Subject: Re: ether collision backoff
Message-ID: <8910132351.AA05712@sparky.Eng.Sun.COM>
Date: 13 Oct 89 23:51:18 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 27


	From: nagle@well.UUCP (John Nagle)
	Subject: Re: ether collision backoff
	Date: 9 Oct 89 16:53:56 GMT
	References: <42686@sgi.sgi.com>

	Back in 1984, somebody at Sun turned down the retransmit delay
	in 4.2BSD's TCP, apparently hoping to "improve performance".
	The resulting mess caused trouble in the Internet for years.

The above is **NOT TRUE**.  For some reason our competitors are always
starting such false rumors.  The truth is that Sun just ported 4.2BSD,
like many other vendors.  Sun just happened to have faster hardware
than most of the others.

The 4.2BSD TCP implementors did not give much thought to the dynamics
of retransmission timers, as everybody should know by now who has read
any recent papers on the subject.  Some work was done at Berkeley to
improve this in the 4.3BSD project, which Sun incorporated into the
SunOS 3.4 release, and even more work was done for the "4.3BSD Tahoe"
release on congestion control that was included in SunOS 4.0.

I could certainly believe that Sun might have subtle bugs in one of its
Ethernet implementations.  These are not intentional.  Please report
any known violations of the spec to the vendor through the appropriate
customer support channels.  Spreading rumors is a bad idea; if it is
broken, get the vendor to fix it!