So is there a way to derive the summary value using the intermediate ones iperf generates, or do I simply need to capture the summary as the final value for the whole transfer?
Thanks
Paul
-----Original Message-----
From: Kevin Gibbs [mailto:kgibbs --at-- ncsa.uiuc.edu]
Sent: Thursday, March 06, 2003 10:13 PM
To: 'iperf-users --at-- dast.nlanr.net'
Subject: Re: Jitter question
Actually the Jitter is not really a number that is calculated per
interval. The Jitter is calculated using the RTP definition and what is
printed at each interval is the value for the whole transfer at that time.
So your calculation is not quite right. The RTP paper is available on the
net, if you want to read why it is a viable calculation and any properties
that it has.
Kevin
On Thu, 6 Mar 2003, paul shields wrote:
> I ran an iperf session on a Windows NT machine (server) from a Sun
> machine(client) that generated the report below. I must misunderstand the
> process of calculating jitter because I don't understand why the final
> jitter is not the same as the average of the sum of the individual report
> intervals.
>
> Are the individual lines worst-case and the final number an average?
>
> The average I get by taking sum (jitter*packets) per line/ total packets =
> ~4.07.
>
> Thanks for any help in understanding
>
> Paul
>
>
>
>
> [ ID] Interval Transfer Bandwidth Jitter Lost/Total
> Datagrams
> [ 32] 0.0- 5.0 sec 63.0 KBytes 103 Kbits/sec 2.605 ms 0/ 3582 (0%)
> [ 32] 5.0-10.0 sec 62.4 KBytes 102 Kbits/sec 3.278 ms 0/ 3549 (0%)
> [ 32] 10.0-15.0 sec 62.1 KBytes 101 Kbits/sec 2.346 ms 0/ 3535 (0%)
> [ 32] 15.0-20.0 sec 61.7 KBytes 101 Kbits/sec 8.112 ms 0/ 3510 (0%)
> [ 32] 0.0-20.0 sec 250 KBytes 102 Kbits/sec 2.937 ms 0/14199 (0%)
>