Packs of scripts are a good idea, however something to monitor the overall 
quailty after the install should always be considered. There is many many many 
things that can go wonk on a voip install from the tos being set wrong on a 
switch to using a voip provider thats too far away (network speaking)
 
Local network switches can be monitored using snmp bundled with cacti or mrtg 
to procude pretty graphical reports, its crude but can give you can idea of 
bandwidth spikes on your lan and collisions,cpu usages etc...
programs like ntop will give you detailed usuage on the packets and break it 
down by program/protocol.
 
 
Wireshark is great as a general tool to see whats happening, when a local lan 
is concerned its always a good tool to get an idea of the bulk/improperly 
configured network traffic, ie you may find some random devices like network 
video cameras spewing large amounts of broadcasts accross your switch.
 

 
________________________________

From: Martin Glazer [mailto:[EMAIL PROTECTED]
Sent: Thu 3/6/2008 12:23 AM
To: Jim Van Meggelen
Cc: Philip Mullis; 'John Lange'; [email protected]
Subject: Re: [on-asterisk] Tools for troubleshooting voice quality





Jim Van Meggelen wrote:
> Ideally, I'd want to come up with some recipies that contain useful tips on
> troubleshooting connectivity and quality problems with the network. How many
> of us run VoIP sets on our customers' networks, and are forced to prove
> network troubles to the customer because the tell us "the network is fine",
> even though there is a lot of evidence to suggest that it is not? Being able
> to produce more detailed data on why we feel the network is "not fine" would
> often be of use. Being able to do that without having to spend a lot of time
> on it, and having access to only the Asterisk server, would make things a
> lot easier, I suspect.

Having a set of scripts or procedures with some metrics that you can
produce would be ideal and extremely valuable.

In the last few months I've had a few client with "perfect" networks
give problems. If a web page or a file takes a few milliseconds longer
to open due to network issues, it is generally not even noticed, but
throw a real time application into the mix and things can go south real
quickly. The problem, as you say, is proving it.

I've found that because we are dealing with UDP packets, this adds to
the difficulties in proving a network issue. At least with TCP, we can
see whether a transmitted packet arrived or was acknowledged, whereas
with UDP, unless you are monitoring both ends (which most of the time
isn't possible) you cannot.

One command that has helped me is netstat - there is a wealth of network
information available. Twice now when I've suspected network problems on
"perfect" networks, I've used netstat to prove that packets were being
dropped/lost and things weren't so happy.

Try netstat -su or netstat -suc for continuous UDP stats.

Martin


Reply via email to