
My test rig setup is as folows;
Below you'll find a number of links to the various code builds that I've tested so far. These tests have been conducted using the Standard SipStone test. This test can be broken down into two main types; AstLinux as the call reciever and AstLinux as the call proxy.
In all cases the "maximum load" and the "operable load" were discovered by trial and error. The "maximum load" measures how many concurrent calls and calls per minute AstLinux can handle before the Asterisk application crashes. A known "feature" of Asterisk is that it ceases to service its system console under heavy load. The "operable load" measures how many concurrent calls and calls per minute the AstLinux can handle whilst still allowing regular use of the system and its console. Each call had a duration of 10000 milliseconds (10 seconds).
In the case of Astlinux being the recipient of all the calls, the trafic has been directed to Asterisk's built in echo test application. This test requires not only that Asterisk routes the call but also generate some RTP media streams for itself thus creating not only potential call volume issues but potential general system resource issues too.
When Asterisk is tested simply as a proxy the anticipated call volume should be much higher due to the routing nature of the proxy. There should be no RTP media or indeed anything else generated by Asterisk. It should simply forward the call according to the initial request.