[Freeswitch-users] Estimating Call Capacity

[Eliot Gable]

Although, FYI, I just benchmarked mod_xml_curl on a separate web app
server from FS with FS on a Dell R710 with their current best
processor option (Intel Xeon X5570 @2.93GHz with 8-cores total) and 32
GB memory. The web app server is less than half the power of the R710.
I maxed the web app server at 300 calls per second (both setting up
and tearing down) and the R710 running FS was 65% idle. No audio was
being proxied through FS, though. If I were running the web app server
on an equivalent R710, they probably would have been on-par with each
other in performance. Extrapolating, I expect that in such a case I
should be able to get at least 650 CPS out of FS, though for
production I would probably limit it to 400 CPS or less so I leave
room for miscellaneous tasks. I maxed out the R710 at over 16,000
simultaneous calls (again, no audio proxying) but the only reason I
couldn't do more was because I hit some sort of thread creation limit
in Linux. There was about 17 GB of memory used for this many calls.
This should give you some ballpark idea of what you can accomplish
with FS.

At some point, I will track down and resolve the thread creation
issue, at which time I believe call limits will be limited either by a
complex combination of available memory, the speed of the processor,
the cost of thread context switching, calls per second setup rate, and
call duration.

--
Eliot Gable

> -----Original Message-----
>
> From: freeswitch-users-bounces at lists.freeswitch.org [mailto:freeswitch-users-bounces at lists.freeswitch.org] On Behalf Of Giovanni Maruzzelli
>
> Sent: Monday, October 26, 2009 4:56 PM
>
> To: freeswitch-users at lists.freeswitch.org
>
> Subject: Re: [Freeswitch-users] Estimating Call Capacity
>
>
>
> On Mon, Oct 26, 2009 at 9:28 PM, Vinuth Madinur
>
> <vinuth.madinur at gmail.com> wrote:
>
> > Here are a few benchmarks that I had stumbled upon.
>
> > http://wiki.voiceworks.pl/display/~pawel/FreeSwitch+performance+on+SUN+x2200+M2
>
>
>
> Please remember NO benchmarks are endorsed by the FS community or
>
> developers, because there are just too many variables, and a simple
>
> figure is just useful for marketing hype, not for real dimensioning.
>
>
>
> You MUST do your own benchmarking, so you get an idea about how to
>
> dimension for your own use case and hardware.
>
>
>
>
>
> > Thanks,
>
> > Vinuth.
>
> >
>
> > On Tue, Oct 27, 2009 at 1:43 AM, Brian West <brian at freeswitch.org> wrote:
>
> >>
>
> >> I highly doubt it... You can wait for someone to post their results
>
> >> but in the end you'll have to do your own load testing because not
>
> >> everyone's numbers will jive with your use case.  Which is the reason
>
> >> the project never posts or endorses a set call count.
>
> >>
>
> >> /b
>
> >>
>
> >> On Oct 26, 2009, at 2:50 PM, Ujjval Karihaloo wrote:
>
> >>
>
> >> > Are there any benchmarking test results available publicly?
>
> >> > ________________________________________
>
> >>
>
> >>
>
> >> _______________________________________________
>
> >> FreeSWITCH-users mailing list
>
> >> FreeSWITCH-users at lists.freeswitch.org
>
> >> http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
>
> >> UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
>
> >> http://www.freeswitch.org
>
> >
>
> >
>
> > _______________________________________________
>
> > FreeSWITCH-users mailing list
>
> > FreeSWITCH-users at lists.freeswitch.org
>
> > http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
>
> > UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
>
> > http://www.freeswitch.org
>
> >
>
> >
>
>
>
>
>
>
>
> --
>
> Sincerely,
>
>
>
> Giovanni Maruzzelli
>
> Cell : +39-347-2665618
>
>
>
> _______________________________________________
>
> FreeSWITCH-users mailing list
>
> FreeSWITCH-users at lists.freeswitch.org
>
> http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
>
> UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
>
> http://www.freeswitch.org