[15153] in North American Network Operators' Group
Re: MTU of the Internet?
daemon@ATHENA.MIT.EDU (John Hawkinson)
Mon Feb 9 01:49:47 1998
From: John Hawkinson <jhawk@bbnplanet.com>
To: smd@clock.org (Sean M. Doran)
Date: Mon, 9 Feb 1998 01:39:36 -0500 (EST)
Cc: nanog@merit.edu
In-Reply-To: <ytafc1eubo.fsf@cesium.clock.org> from "Sean M. Doran" at Feb 8, 98 03:05:15 pm
> Ebone will be deploying code that does RED in CEF in
> approximately nine hours. I expect BC and I will be
> playing with RED most of this week, seeing how the Cisco
> implementation does on congested high-speed and sometimes
> high-delay interfaces where traffic is very heavily aggregated.
>
> I am not sure anybody else has done this yet (jhawk?)
> although there is some positive experience with RED on
> slow interfaces.
Some. A few links, only one of which sees enough drops for RED to make
eny difference at all. I've been remiss :-(, though oddly enough I've
spent some cycles with it last week, and have some more slotted this
week.
If people are interested this link (unidirectionally) is doing RED'd:
8 paix.bbnplanet.net (4.0.1.50) 80.247 ms 80.399 ms 80.463 ms
9 su-bfr.bbnplanet.net (4.0.1.49) 82.032 ms 81.574 ms 80.578 ms
BTW, if you want to have a numbers-based discussion of your
results, I suggest big-internet.
I suppose we need to come up with some good guidelines for tweaking
parameters...
--jhawk
