[Asterisk-Users] Anyone using SipP to produce RTP load?

[tim panton]

On 29 Jun 2005, at 04:51, Matthew Boehm wrote:


> Hey gang,
>  I've been able to use sipp to produce some call volume on our  
> asterisk
> server. The server has no problems handling 50 simul calls. But  
> then again,
> no RTP is being done. I tried to use the rtp echo ability of sipp  
> but that
> doesn't seem to work right.
>  I also setup a fake number in asterisk that when called by sipp,  
> would dial
> another number via PRI, hoping that some 729 conversion would occur.
> Nothing. I was able to pump 10 simul calls that went this path:
>
>   sipp -> asterisk -> pri -> telco ->pri ->asterisk
>
> ..and still no 729 usage or any other discernable load on the server.
>
> Can anyone offer suggestion on how to really simulate calls (using  
> sipp or
> other tester) to asterisk to verify its ability to process X calls?
>
> I know someone out there has done this, but forget who it was.
>

I think you mean Signate.
I saw a presentation at Astrcon .
They call the milliwatt generator to fill the RTP stream.

They were getting 122 passthrough ulaw calls on a 'stock' pc.
If I remember right the benchmark scripts and methodology are
available.


If you are looking to benchmark that 4 way 500Mhz box of yours
I'd be _very_ interested in the results with varying numbers of CPUs.
Signate were saying that the limiting factor (with ulaw passthrough)
is the PC architecture (bus and interrupt structure) not the CPU.

I've done a _tiny_ experiment myself. I found that a single 729->alaw- 
 >PRI
call uses less than 10% of the CPU on a 1Ghz nemiah

Tim.


>
> Thanks,
> Matthew
>