The test environment was an Intel quad core machine with 8 GB and FreeSWITCH Version 1.0 installed. The load was generated by SIPP running on a different machine.
On this standard server, and just a little bit of tweaking, with 2000 calls in 50 different conference rooms with each conference being recorded to a file - the call quality was still ok and cpu was 50% idle.
Here are the SIPP result screens
![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgucrf6amJ4o7H-N1xUezjOqbQM3M8GFHYZsZVLZyhNhX8K42NkEWlOaZWPwdVqtyB3ynsk65YYhPPMaPEFahJ26iFzk9y-zlJqTkWr_XMmmSy3z4GdyR3wzMBX8GodwSJcG1_x2GKM-Tc/s320/mod_conference_sipp_scenario.png)
![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEij0kF8kwV-gtSeTZueknNuodkCeMu2KmCmly3Vq0JJAJmpJb8zcRW8DKcGNqjutkSHPnYkiZlNC8WUekSyEkRsTqnn4SoyMrtNNxxvh6iUAFYkNFUtt5hK09z9OAnY3wRI-wds6xCrhgc/s320/mod_conference_sipp_statistics.png)
Each call lasted about 4 and a half minutes with more or less 10% of that time prerecorded alaw sound was streamed by SIPP at different intervals throughout the call. 2002 calls were connected at peak, with 2000 generted by SIPP and 2 calls were placed manually to test the call quality.
I would say the results are quite impressive, what do you think ?