RE: Stream BW support in Iperf
Thanks for your response Ian.
I totally agree with what you are saying, but we get a better
performance per stream with Netperf.
We are trying to analyze Iperf Anomaly for smaller buffer sizes of 64
bytes.
We get around 89Mbits/stream for Iperf and about 230Mbits/stream for
Netperf.
So more than the TCP protocol implementation inside the OS, is there
some kind of basic Iperf implementation issue which causes this
difference when compared to Netperf?
They say that Netperf doesn't implement traffic shaping while Iperf
implements traffic shaping. Will this be a reason? If that's the reason
is there a way to turn it off?
Your help is greatly appreciated.
Thanks,
Samson
-----Original Message-----
From: owner-iperf-users --at-- dast.nlanr.net
[mailto:owner-iperf-users --at-- dast.nlanr.net] On Behalf Of Ian McDonald
Sent: Tuesday, September 19, 2006 4:13 PM
To: iperf-users --at-- dast.nlanr.net
Subject: Re: Stream BW support in Iperf
> Does anyone know why we need multiple streams to max out the available
> Bandwidth?
>
> Why can't a single stream support high bandwidths?
>
Because TCP can't use all the available bandwidth due to congestion
control. There have been many attempts to improve TCP but the thing is
that loss comes in many forms, and RTT has a big impact on algorithms
too.
Try googling for TCP congestion control. You can also see the effects
by taking iperf on linux and changing the congestion control and use
netem or dummynet to do some traffic shaping. If you go to my website
you can even download versions which allow you to do this without
altering sysctls.
Ian
--
Ian McDonald
Web: http://wand.net.nz/~iam4
Blog: http://imcdnzl.blogspot.com
WAND Network Research Group
Department of Computer Science
University of Waikato
New Zealand