[95986] in North American Network Operators' Group
Re: Thoughts on increasing MTUs on the internet
daemon@ATHENA.MIT.EDU (Joe Loiacono)
Thu Apr 12 16:49:10 2007
In-Reply-To: <Pine.LNX.4.64.0704122203500.28703@uplift.swm.pp.se>
To: Mikael Abrahamsson <swmike@swm.pp.se>
Cc: NANOG list <nanog@merit.edu>, owner-nanog@merit.edu
From: Joe Loiacono <jloiacon@csc.com>
Date: Thu, 12 Apr 2007 16:31:48 -0400
Errors-To: owner-nanog@merit.edu
This is a multipart message in MIME format.
--=_alternative 00709236852572BB_=
Content-Type: text/plain; charset="US-ASCII"
owner-nanog@merit.edu wrote on 04/12/2007 04:05:43 PM:
>
> On Thu, 12 Apr 2007, Joe Loiacono wrote:
>
> > Large MTUs enable significant throughput performance enhancements for
> > large data transfers over long round-trip times (RTTs.) The original
>
> This is solved by increasing TCP window size, it doesn't depend very
much
> on MTU.
Window size is of course critical, but it turns out that MTU also impacts
rates (as much as 33%, see below):
MSS 0.7
Rate = ----- * -------
RTT (P)**0.5
MSS = Maximum Segment Size
RTT = Round Trip Time
P = packet loss
Mathis, et. al. have 'verified the model through both simulation and live
Internet measurements.'
Also (http://www.aarnet.edu.au/engineering/networkdesign/mtu/why.html):
"This is shown to be the case in Anand and Hartner's "TCP/IP Network Stack
Performance in Linux Kernel 2.4 and 2.5" in Proceedings of the Ottawa
Linux Symposium, 2002. Their experience was that a machine using a 1500
byte MTU could only reach 750Mbps whereas the same machine configured with
9000 byte MTUs handsomely reached 1Gbps."
AARnet - Australia's Academic and Research Network
>
> Larger MTU is better for devices that for instance do per-packet
> interrupting, like most endsystems probably do. It doesn't increase
> long-RTT transfer performance per se (unless you have high packetloss
> because you'll slow-start more efficiently).
>
> --
> Mikael Abrahamsson email: swmike@swm.pp.se
--=_alternative 00709236852572BB_=
Content-Type: text/html; charset="US-ASCII"
<br><font size=2><tt>owner-nanog@merit.edu wrote on 04/12/2007 04:05:43
PM:<br>
<br>
> <br>
> On Thu, 12 Apr 2007, Joe Loiacono wrote:<br>
> <br>
> > Large MTUs enable significant throughput performance enhancements
for<br>
> > large data transfers over long round-trip times (RTTs.) The original<br>
> <br>
> This is solved by increasing TCP window size, it doesn't depend very
much <br>
> on MTU.</tt></font>
<br>
<br><font size=2><tt>Window size is of course critical, but it turns out
that MTU also impacts rates (as much as 33%, see below):</tt></font>
<br>
<br><font size=2><tt> MSS 0.7</tt></font>
<br><font size=2><tt>Rate = ----- * -------</tt></font>
<br><font size=2><tt> RTT (P)**0.5</tt></font>
<br>
<br><font size=2><tt>MSS = Maximum Segment Size</tt></font>
<br><font size=2><tt>RTT = Round Trip Time</tt></font>
<br><font size=2><tt>P = packet loss</tt></font>
<br>
<br><font size=2><tt>Mathis, et. al. have 'verified the model through both
simulation and live Internet measurements.'</tt></font>
<br>
<br><font size=2><tt>Also (http://www.aarnet.edu.au/engineering/networkdesign/mtu/why.html):
</tt></font>
<br>
<br><font size=2><tt>"This is shown to be the case in Anand and Hartner's
"TCP/IP Network Stack Performance in Linux Kernel 2.4 and 2.5"
in Proceedings of the Ottawa Linux Symposium, 2002. Their experience was
that a machine using a 1500 byte MTU could only reach 750Mbps whereas the
same machine configured with 9000 byte MTUs handsomely reached 1Gbps."</tt></font>
<br>
<br><font size=2><tt>AARnet - Australia's Academic and Research Network</tt></font>
<br><font size=2><tt><br>
> <br>
> Larger MTU is better for devices that for instance do per-packet <br>
> interrupting, like most endsystems probably do. It doesn't increase
<br>
> long-RTT transfer performance per se (unless you have high packetloss
<br>
> because you'll slow-start more efficiently).<br>
> <br>
> -- <br>
> Mikael Abrahamsson email: swmike@swm.pp.se<br>
</tt></font>
--=_alternative 00709236852572BB_=--
