Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[mininet-discuss] two consecutive iperf measures: very different output values

Nicholas Bastin nick.bastin at gmail.com
Wed Oct 28 12:39:28 PDT 2015


On Tue, Oct 27, 2015 at 11:28 AM, Massimo Canonico <mex at di.unipmn.it> wrote:

> then I run iperf 10 times in each of my 5 run and I compute the average.
>
> First run with 10 "iperf" run --> Average : 578Mbps
> Second run with 10 "iperf" run --> Average : 590Mbps
> Third run with 10 "iperf" run --> Average : 594Mbps
> Fourth run with 10 "iperf" run --> Average : 586Mbps
> Fifth run with 10 "iperf" run --> Average : 592Mbps
>
> Now the questions are at least two:
> 1. since the network is so simple, I should see something close to 800Mbps
> but as you can see, the iperf outputs are quite far away from the reference
> value. Where does mininet lose 200Mbps?
>

Whether you're likely to get 800Mbps involves a lot of factors, not least
of which are the TCP parameters on your hosts (window size, scaling, etc.),
whether shaping is causing uneven queuing or any reordering or drops.


> 2. is it acceptable that the values range is [578, 594] in such a simple
> and unloaded network?
>

That variance is pretty reasonable, without knowing more about your
hardware.

--
Nick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/mininet-discuss/attachments/20151028/3b0cb0b3/attachment.html>


More information about the mininet-discuss mailing list