Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 28 Jan 1997 17:40:18 -0600 (CST)
From:      jpt@msc.edu (Joseph Thomas)
To:        danny@panda.hilink.com.au (Daniel O'Callaghan)
Cc:        shovey@buffnet.net, robert@nanguo.chalmers.com.au, freebsd-isp@freebsd.org, hackers@freebsd.org
Subject:   Re: RFC 1323 default settings (was Re: progress report on connection problems)
Message-ID:  <199701282340.RAA10860@ww.msc.edu>
In-Reply-To: <Pine.BSF.3.91.970129092110.13981t-100000@panda.hilink.com.au> from "Daniel O'Callaghan" at Jan 29, 97 09:40:21 am

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> 
> 
> On Tue, 28 Jan 1997, Steve wrote:
> 
>  [ Robert writes about his problem with FreeBSD machines and his Annex]
> > 
> > I had similar problems using annexes as term servers with user - and have
> > posted numerous times that this problem only exists with freebsd  - sco,
> > linux, etc doesnt have trouble - only every time I post it I get bashed
> > about the head and lectured on freebsd having perfect tcp/ip and
> > everything else in the world having faulty tcp/ip.
> 
> Well, turns out that a Linux 2.0 box also stalled on this.  However, 
> methinks it might be time to raise the issue of default settings for RFC 
> 1323 extensions in FreeBSD boxes.
> 
> Since RFC 1323 deals with long fat pipes, which very few of us have, it
> would make sense to turn the extensions off in the shipped /etc/sysconfig. 
> Those people who are communicating with TCP between the two ends of a 1.5
> Mbps satellite link, or an intercontinental 45 Mbps link, probably know
> who they are and can turn the extensions on.  I know that some people in
> the continental USA can claim a 45 Mbps path to their favourite ftp site 4
> states away, but surely those paths are used by others, reducing the
> 'fatness' of the pipe for RFC 1323 purposes.  Local area ATM might be 
> 'fat' but generally is not 'long' enough to cause the problems with RFC 
> 1323 addresses.

	As a data point - running a local-area ATM with "out of the box"
parameters (for 2.2 this looks to be 16K windows with no-scaling), I get
60 KB/s out of the box vs 3.0-3.5 MB/s into the box, [notice the really
bad discrepancy] via ftp. With larger windows (60KB), I can get in the
range of 3.5-4.0 MB/s [either 'put xxx /dev/null' or 'get xxx /dev/null'
so local disk access is somewhat unrelated. That is, the numbers don't
vary much if I'm sending from local disk or receiving to /dev/null.]

	Using ttcp (tcp user application, memory to memory), I've transmitted
close to 70 Mb/s, in the "local-area". I'm not sure that getting twice
the throughput counts as being 'not long enough'.

[I'm simply providing this as a data point for the discussion, not attempting
or interested in arguing for or against either side.]

> 
> Leaving the extensions on by default causes much grief for people with old
> Annexes, prevents people whose ISPs use Annexes from reading FreeBSD box
> web pages or sending mail to FreeBSD boxes, and generates enormous amounts
> of traffic on the FreeBSD mailing lists. 
> 
> regards,
> 
> Danny
> 


-- 
Joseph Thomas                           E/Mail:  jpt@msc.edu
Minnesota Supercomputer Center, Inc.    	 jpt@magic.net
1200 Washington Ave So.			Tel:	 +1 612 337 3558
Minneapolis, MN     55415-1227          FAX:     +1 612 337 3400

	You cannot see what I see because you see what you see.
	You cannot know what I know because you know what you know.
				"Mostly Harmless" - Douglas Adams



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199701282340.RAA10860>