Re: Iperf TCP throughput measurement result


On 5/26/06, Jin Fan <jfan --at-- cwlab.com> wrote:
Hi,

We have been trying to eliminate the loss with attenuators when we test those radios on the bench.

The tcp window size is set to 65k.  Again, the distance between radios are very short.  The size of congestion window shouldn't be an issue.

Jin

________________________________

From: owner-iperf-users --at-- dast.nlanr.net on behalf of Ian McDonald
Sent: Mon 5/15/2006 5:49 PM
To: iperf-users --at-- dast.nlanr.net
Subject: Re: Iperf TCP throughput measurement result



>
> Any comments on this issue will be appreciated.
>
Often with TCP one stream cannot use all the bandwidth available
particularly if there is loss - which is more likely with wireless.
What TCP variant are you using as some cope with loss better than
others?

Our research group has a TCP calculator available here you might be
interested in:
http://wand.net.nz/~perry/max_download.php

Ian

Go and check out the calculator as I suggested...

I put in 0.3% loss, 3 msec RTT, 64 K window, 300 Mbit upstream and I
get 60 Mbit maximum for one TCP flow...

This is just based on the TCP equations. It is a common misconception
that you can always drive a link at it's full speed with TCP.

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



Other Mailing lists | Author Index | Date Index | Subject Index | Thread Index