<br><br><div class="gmail_quote">On Wed, Jun 17, 2009 at 7:41 PM, Steve Totaro <span dir="ltr"><<a href="mailto:stotaro@totarotechnologies.com">stotaro@totarotechnologies.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><div></div><div class="h5"><br><br><div class="gmail_quote">On Wed, Jun 17, 2009 at 3:18 PM, John Todd <span dir="ltr"><<a href="mailto:jtodd@digium.com" target="_blank">jtodd@digium.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><div></div><div><br>
On Jun 17, 2009, at 8:16 AM, Steve Totaro wrote:<br>
<br>
> Hi,<br>
><br>
> Quick question to the real world.<br>
><br>
> Approx what specs would I need on server to handle 95 ZAP or Dahdi -<br>
> > SIP gateway using G729 on the SIP to carrier side (nothing else,<br>
> just media conversion)?<br>
><br>
> Does the latest Asterisk/DAHDI significantly improve these numbers<br>
> over say, Asterisk 1.2.X?<br>
><br>
> Sure, there is plenty to read but nothing I could find quickly on my<br>
> exact needs that was clear and I want to be fairly sure before<br>
> ordering a server.<br>
><br>
> Obviously load avg has something to do with it but CPU and mem seems<br>
> to be the biggest factors.<br>
><br>
> --<br>
> Thanks,<br>
> Steve Totaro<br>
> +18887771888 (Toll Free)<br>
> +12409381212 (Cell)<br>
> +12024369784 (Skype)<br>
<br>
<br>
</div></div>[Digium hat off, ITSP (previous employer) hat on.]<br>
<br>
Not speaking for Digium on this one, but speaking from personal<br>
experience at another company.<br>
<br>
We could get 100 G.729 channels in a 2x3ghz P4 machine with plenty of<br>
CPU room to spare, 4+ years ago, and that rule of thumb served us<br>
well. This was for SIP (G.711<->G.729) but I can't imagine that DAHDI<br>
or Asterisk takes significantly more or less horsepower for this task<br>
now for the base transcoding load.<br>
<br>
JT<br>
<br>
---<br>
<font color="#888888">John Todd <a href="mailto:email%3Ajtodd@digium.com" target="_blank">email:jtodd@digium.com</a><br>
Digium, Inc. | Asterisk Open Source Community Director<br>
445 Jan Davis Drive NW - Huntsville AL 35806 - USA<br>
direct: +1-256-428-6083 <a href="http://www.digium.com/" target="_blank">http://www.digium.com/</a><br>
</font><div><div></div><div><br>
<br>
<br></div></div></blockquote></div><br></div></div>John,<br><br>Thanks for the info.<br><br>Can't go wrong with these in a cluster if you are correct. <a href="http://www.surpluscomputers.com/348663/hp-dl140-proliant-dual-xeon.html" target="_blank">http://www.surpluscomputers.com/348663/hp-dl140-proliant-dual-xeon.html</a><br>
<br>I used to have a garden of seven (too small to be a farm) of Asterisk boxen that were simple PRI to SIP gateways doing ulaw (no transcoding as far as codec), and nothing else running, just a few lines in each conf file.<br>
<br>They would run at ~65% CPU utilization with 95 channels in use on each box.<br><br>This was the Asterisk, Zaptel, Libpri 1.2.X flavor.<br><br>I wonder if anyone else has input on the CPU utilization of of a simple PSTN <-> VoIP gateway based on Asterisk?<div>
<div></div><div class="h5"><br clear="all">
</div></div></blockquote><div> </div></div>I wonder how much of my observed 65% CPU utilization was because of software echo cancellation?<br><br>If someone knows, I would appreciate any input. If not, I will follow-up with my own testing in this thread.<br>
<br>-- <br>Thanks,<br>Steve Totaro <br>+18887771888 (Toll Free)<br>+12409381212 (Cell)<br>+12024369784 (Skype)<br>