<p>There is a tradeoff between cps and audio quality and timing accuracy.</p>
<p>Try -vm -nocal if you want to mimic 1.0.4 we don't discuss performance here.</p>
<p><blockquote type="cite">On Jan 31, 2010 8:56 AM, "Robin Vleij" <<a href="mailto:robin@swip.net" target="_blank">robin@swip.net</a>> wrote:<br><br>On 1/31/10 2:57 AM, Michael Jerris wrote:<br>
<br>
Hi Mike!<br>
<p><font color="#500050"><br>> For clarification, is it correct that your getting worse numbers for<br>> sustainable cps on SVN the...</font></p>Yes, 1.04 offered a lot better performance in my test scenario, using<br>
the same call scripts. I agree that the scenario isn't really realistic,<br>
I should test with at least 30 second calls.<br>
<p><font color="#500050"><br>> numbers you would get with bypass_media=true instead of proxy_media=true<br></font></p>Thought about the same, and funnily enough that didn't help. But in the<br>
long list of things I tried I don't remember if I tried it on 1.04,<br>
without media proxying. I expect a much higher performance just<br>
bypassing the media. I'll try again tomorrow on 1.04 and post the results.<br>
<p><font color="#500050"><br>> and with neither setting set as well. Also, make sure your logging level<br>> is low and try putting...</font></p>I check disk performance using iostat and they're on 2% usage. It's a<br>
10k rpm sas disk we're writing to, with ram cache on the raid<br>
controller. Also -nosql didn't have any influence at all on performance.<br>
<p><font color="#500050"><br>> that length of call matters very little. In proxy media or normal mode,<br>> the performance of the ...</font></p>Exactly. I'll adjust the lengt of the call to 30 or 60 seconds and then<br>
try to see how many calls we can have. That's in the end what matters<br>
really and not cps on it's own. Have to create a 60 second pcap first. :)<br>
<p><font color="#500050"><br>> calls). Some other tips. While this extension may be trivial, what else<br>> is there in your dialpl...</font></p>Nothing, it's a single entry dialplan and that entry is thus on the top.<br>
<p><font color="#500050"><br>> cause a significant impact. Do you have any of the presence features<br></font></p>All presence features disabled.<br>
<p><font color="#500050"><br>> enabled? These do significantly impact call handling performance even if<br>> your sipp scenarios do...</font></p>Yep. :)<br>
<br>
Thanks for the input so far!<br>
<br>
Maybe after I get some final results I put it on the wiki for future<br>
reference to load tests?<br>
<font color="#888888"><br>
/robin<br>
</font><p><font color="#500050"><br>_______________________________________________<br>FreeSWITCH-users mailing list<br>FreeSWITCH-users@list...</font></p></blockquote></p>