[3425] in linux-net channel archive
Re: shaper or whatever
daemon@ATHENA.MIT.EDU (Alan Cox)
Sun Jun 23 15:17:23 1996
From: alan@lxorguk.ukuu.org.uk (Alan Cox)
To: schenk@cs.toronto.edu (Eric Schenk)
Date: Sun, 23 Jun 1996 17:10:59 +0100 (BST)
Cc: dennis@etinc.com, linux-net@vger.rutgers.edu
In-Reply-To: <96Jun22.144239edt.15395@dvp.cs.toronto.edu> from "Eric Schenk" at Jun 22, 96 02:42:37 pm
> Sure. But combined congestion window should not be confused with the
> maximum window available on the combined TCP's open across a link.
> In the long run the combined congestion windows should reflect the
> real bandwidth of the link. In reality TCP is good at reacting to
> increased congestion, but not very good at quickly taking advantage
> of extra bandwidth, so the combined congestion windows could often
> be smaller than the real avialable bandwidth.
There is a second big problem to note. With a 64K link and a big web server
(yes its the web messing things around again) you often don't get enough time
to build accurate rtt and congestion window estimates. There is a second case
you should never get too unless you have a bad ISP at which point the per
socket congestion window for each connection over a very loaded line falls below
one. That case is not pretty...
Alan
