From owner-freebsd-hackers  Sat Mar 31 12:56:13 2001
Delivered-To: freebsd-hackers@freebsd.org
Received: from info.iet.unipi.it (info.iet.unipi.it [131.114.9.184])
	by hub.freebsd.org (Postfix) with ESMTP id A999437B719
	for <FreeBSD-hackers@FreeBSD.ORG>; Sat, 31 Mar 2001 12:56:09 -0800 (PST)
	(envelope-from luigi@info.iet.unipi.it)
Received: (from luigi@localhost)
	by info.iet.unipi.it (8.9.3/8.9.3) id WAA33314;
	Sat, 31 Mar 2001 22:55:22 +0200 (CEST)
	(envelope-from luigi)
From: Luigi Rizzo <luigi@info.iet.unipi.it>
Message-Id: <200103312055.WAA33314@info.iet.unipi.it>
Subject: Re: unexpected peaks in dummynet;scheduling oddity?
In-Reply-To: <20010331111427.A3209@devil.rrze.uni-erlangen.de> from Jochen Kaiser
 at "Mar 31, 2001 11:14:27 am"
To: Jochen Kaiser <Jochen.Kaiser@rrze.uni-erlangen.de>
Date: Sat, 31 Mar 2001 22:55:22 +0200 (CEST)
Cc: FreeBSD-hackers@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL61 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Hi,
the packet scheduler is invoked by dummynet at every packet arrival
and then at multiple of 1/HZ intervals. If you use the default
kernel setting, HZ=100 so you have most times rounded to multiples
of 10ms. I have been running my kernels with HZ=1000 for the past
4-5 years so the resolution is much better -- 1ms, which is  appropriate
for links up to a few megabits/s

I think the above should explain the results you see.

	cheers
	luigi

> Hello,
> 
> I did some measurements using Dummynet. It was just to see how
> it works. Tests are made with 0,7MBit stream (each 429Bytes Packets),
> 5,10,15,25,50 and 75 MBit.
> 
> I tried a delay of 10ms for the testing connection. 
> The tests were done with a Smartbits6000 with the capability to
> record the difference between sending and receiving a packet.
> Each test had a 8192 packets.
> 
> 
> The result show a strange behaviour, depending on the data rate.
> 
>    _
>   | |    vs.    /\
>                                                                  
> 0.7MBit - 15MBit: diffuse, fairly hig variance from 9.5 - 10.3ms 
> 
> 25 MBit: between
> 
> 50MBit - 75MBit: very sharp, extreme precise, delivers a nearly 10ms
> 
> (variance not calculated yet ... to be done ..)
> 
> I am now a bit confused. Is it a matter of scheduling?
> Maybe too much fairness for such tasks :)
> 
> Or is it some kind of prediction in dummynet which favours 
> high data rates. I think I've seen something like that for packet
> loss in dummynet.
> 
> 
> Any opionions appreciated. Maybe I did something wrong 
> and forget to set some sysctl properly.
> 
> with kind regards,
> Jochen Kaiser
> --
> Jochen Kaiser                kind@IRCNET, phone +49 9131 85-28134
> Network Administration  mailto:jochen.kaiser@rrze.uni-erlangen.de
> Regionales Rechenzentrum Universitaet Erlangen-Nuernberg, Germany
> GPG public key: http://www.uni-erlangen.de/~unrza2/public_key.txt
> 


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message