[asterisk-users] Fw: Stress testing Asterisk

[Matias Banchoff]

El 22/05/13 12:25, Paul Belanger escribió:
> On 13-05-22 10:02 AM, Tommy Cooper wrote:
>>  From the little experience I have I do not think that that is a good 
>> way of testing the quality of voice. SIP only initiates and 
>> eventually terminates the call, once that the call is connected, SIP 
>> and therefore Asterisk are no longer involved. Once the call is 
>> connected it is assigned to a trapsport layer protocol such as RTP. 
>> RTP is the actual protocol that delivers the voice call between 
>> endpoints. I believe that the setup of your network, QoS, codecs 
>> etc... determine the voice quality of your system.
>>
>>
>> ----- Forwarded Message -----
>> From: Mitul Limbani <mitul at enterux.in>
>> To: Tommy Cooper <tomcooper83 at yahoo.com>; Asterisk Users Mailing List 
>> - Non-Commercial Discussion <asterisk-users at lists.digium.com>
>> Sent: Wednesday, May 22, 2013 3:23 PM
>> Subject: Re: [asterisk-users] Stress testing Asterisk
>>
>>
>>
>> I have a question here.
>>
>> How can we test the quality of voice upon increasing the call load?
>>
>> Can we try passing a voice file using sipp and record the same in 
>> dial plan record application ? Is this reliable enough to simulate 
>> near real world scenario?
>>
>
> Once upon a time, we set out to create exactly this for testing 
> asterisk.  Our goal would have been to run the test every week, 
> comparing the results from the previous week, to make sure asterisk's 
> performance was not getting worse as new commits happened.
>
> We came up with the idea of loading testing asterisk using SIPp or 
> some other dialer, then determining at what point asterisk would start 
> failing (performance).  We decided the point of failure was quality of 
> audio, since it is usually the first thing to go (even though call 
> control still works).
>
> It took a while, but with the help of Leif, we found a tool to analyse 
> audio streams (using MOS score[1]).  Basically, you take the original 
> audio file, play it across the network, then record the other side. 
> Then, comparing the two files via Aqua, you get your MOS score.
>
> If the score was less then x, you knew asterisk was hitting a 
> performance limit.  Track that over time and concurrent calls, you 
> have your metrics.
>
> [1] http://www.sevana.fi/aqua.php
>
Hi!
   I haven't used it, but there is a quality test algorithm provided by 
ITU.

http://stackoverflow.com/questions/2329403/how-to-start-a-voice-quality-pesq-test
http://en.wikipedia.org/wiki/PESQ
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=6043771&queryText%3DDevelopment+of+a+Speech+Quality+Monitoring+Tool+based+on+ITU-T+P.862



-----
CeSPI 
Centro Superior para el Procesamiento de la Información

Universidad Nacional de La Plata
-------------------------------------------------------------------------------
Proteja el Medioambiente. No imprima este mail si no es absolutamente necesario