[c-nsp] Opensource tool to measure Jitter for VoIP

[Phil Vandry]

On Mon, Jun 08, 2009 at 07:46:06AM -0500, Bryan Campbell wrote:
> You cannot measure VOIP (sip) jitter using ICMP tools.  You will only

s/sip/RTP/

[snip using Wireshark VoIP analysis]
> If you can't find jitter in this manner, it cannot be found.  If it
> cannot be found, it doesn't exist.

This will be true as long as you are monitoring close to the receiving
end. Otherwise you will miss jitter that is introduced by the network
beyond your monitoring point.

This means you may want to have two monitoring points for bidirectional
voice traffic: one close to each receiving end.

On Mon, Jun 08, 2009 at 10:06:21AM -0400, Eric Van Tol wrote:
> What are the there legal ramifications to this?  While I like to think
> that "it's my network, I'll do what I want to measure its performance",

You could avoid problems by capturing (Wireshark or tcpdump) using a
limited snapshot length. You do not need the payload to perform jitter
analysis so "tcpdump -s 50" might be safer (14 bytes Ethernet + 20 IP +
8 UDP + 8 RTP header).

-Phil