From owner-freebsd-questions  Thu Nov 13 22:29:48 1997
Return-Path: <owner-freebsd-questions>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id WAA01175
          for questions-outgoing; Thu, 13 Nov 1997 22:29:48 -0800 (PST)
          (envelope-from owner-freebsd-questions)
Received: from implode.root.com (implode.root.com [198.145.90.17])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id WAA01170
          for <questions@freebsd.org>; Thu, 13 Nov 1997 22:29:45 -0800 (PST)
          (envelope-from root@implode.root.com)
Received: from implode.root.com (localhost [127.0.0.1])
	by implode.root.com (8.8.5/8.8.5) with ESMTP id WAA29131;
	Thu, 13 Nov 1997 22:29:52 -0800 (PST)
Message-Id: <199711140629.WAA29131@implode.root.com>
To: "Lee Crites (AEI)" <leec@adam.adonai.net>
cc: questions@freebsd.org
Subject: Re: Walnut Creek 
In-reply-to: Your message of "Thu, 13 Nov 1997 19:16:14 CST."
             <Pine.BSF.3.95.971113191147.12421A-100000@adam.adonai.net> 
From: David Greenman <dg@root.com>
Reply-To: dg@root.com
Date: Thu, 13 Nov 1997 22:29:52 -0800
Sender: owner-freebsd-questions@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>On Wed, 12 Nov 1997, David Greenman wrote:
>
>=>>Ah, thanks -- but you didn't say what version they are running. 
>=>
>=>   2.2-stable with local changes. Don't ask for the local changes.
>
>If I remember correctly, the "local changes" were about using a
>dumptruck load of ram.  I'm thinking it was quite a bit more than
>the normal person would ever use, so it wouldn't matter to most
>of us anyway.
>
>Just out of idle curiosity, are there other "local changes"
>besides the memory one?
>
>My main reason for asking is because I use www.cdrom.com as an
>example of how kick-butt FreeBSD is.  If that site is better
>because of a bunch of "local changes," then I need another
>example...

   I thought I said "Don't ask...". :-)
   Don't worry too much, the only significant difference in the code running
on ftp.cdrom.com is that TCP delayed acks have been disabled. This is a
special-case performance improvement that is tailored to our very special
circumstances. I haven't made this a tunable option in FreeBSD because I
want to fix the problem correctly by changing the way that some of the TCP
timer mechanisms work (esp wrt delayed acks). This and other TCP scalabilty
improvements are on my short term TODO list and I hope to have them completed
prior to the 3.0 release.
   Other than the above, the other changes we're running with are selective
retrofits of improvements that I've made to the -current FreeBSD sources.
Some of these might be brought into 2.2-stable now that we've got the 2.2.5
release out of the way, but I haven't made any specific decisions yet.

-DG

David Greenman
Core-team/Principal Architect, The FreeBSD Project