Wireshark-users: Re: [Wireshark-users] VoIP analysis and assessment

From: "Chris Swinney" <swin@xxxxxxxxxxxxx>
Date: Wed, 4 Oct 2006 19:59:06 +0100
Many thanks for the direction. Have I read the Wireshark docs right in
that it can reconstruct the audio of VoIP call from the packets
captured?

I will firstly attempt to run simultaneous captures at both ends,
inserting a 100Mbit hub between the router and PBX switch with PC's
running Wireshark connected to the hub (can't afford a network probe).
This will hopefully capture all traffic passing to the router including
that destined to be sent across the VPN without causing to much of a
skew to the data (due to the non duplex nature of the hub). If I can
then reconstruct the audio I might have a better understanding of what
the user is experiencing. 

I will try an repeat the exercise with a software based VoIP product
that can be run from the two capture PC's assuming they will be up to
the task (any suggestions?), or other VoIP hardware that can be used to
connect direct without the need for the PBX. I hope this will at least
give an indication of where the fault lies.

However, I am beginning to suspect the PBX switches. We ran a simple
test by disconnecting the rest of the data networks which made little/no
difference. We are well within our bandwidth budgetary constraints (150
kbps) and the ISP latency between sites is only 45ms.

If Wireshark isn't really the tool to get accurate quality scores, has
anyone got any suggestions for a tool that might be better suited to the
task - without breaking the bank?

 
Thanks,
 
Chris
-----Original Message-----
From: Jaap Keuter [mailto:jaap.keuter@xxxxxxxxx] 
Sent: 28 September 2006 07:08
To: Community support list for Wireshark
Subject: Re: [Wireshark-users] VoIP analysis and assessment

Hi,

[CLIP elaborate description]

I think Wireshark is of limited use here. Things like MOS determination
is
not in it's realm.
But it can be helpfull. I would go for test calls. Get down to the ends
where the "buzz" problem is happening and from there call into an
autoanswering upstream extension connected to a tone source. Capture the
RTP traffic with Wireshark once determined that the "buzz" occurs. Save
the payload of this capture and see what's made of it. If the buzzing is
in there, you've got issues with the PSTN/VoIP codecs, if it's not the
jitterbuffer in the terminals may be wacky.

Just my E0.02
Jaap