[Click] Measuring

Laercio Cruvinel Junior lcruvinel at inesc-0.tagus.ist.utl.pt
Mon Mar 5 19:00:28 EST 2007


I would recommend the use of external tools for capturing statistics on
packets that allow extracting the data for measuring delays and jitter of
UDP flows, namely rude/crude/ethereal. Using the Socket element a flow can
be directed to a crude listener. The alternative is developing customized
C++ click elements.
If you use a timed source with constant packet size, measures are more
meaningful. In this case, there is no need to use only one machine for
getting the jitter, because the differences between packets departure time
are always known and the differences between packets arrival time are got
from dumps on destination machine.
Regards,

Laercio Cruvinel

> From: "Egi, Norbert" <n.egi at lancaster.ac.uk>
> Hi,
>
> Is there any way to measure packet delay and jitter with Click?
>
> I'm using Click's FastUDPSource to generate packets on a machine, which
> packets are forwarded through another machine and are counted on a third
> with the AverageCounter element. This works quite good for measuring the
> packet forwarding rate and loss of the system in the middle, but I would
> also like to measure the delay and jitter of the packets. I guess to do
> that the traffic should be generated and counted on the same machine,
> which isn't a problem, but what elements could be used to do that? I found
> the StoreTimestamp element, which stores the packet timestamps in packet
> data, but couldn't find any element that is able to get this timestamp out
> of the packet after it has been received.
>
> Any suggestion how to do it if it's possible at all?
>
> Many thanks,
> Norbert
>
>
>
> ------------------------------
>
> _______________________________________________
> click mailing list
> click at amsterdam.lcs.mit.edu
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>
>
> End of click Digest, Vol 45, Issue 3
> ************************************
>




More information about the click mailing list