[Freeswitch-users] Max of 170 channels in the conference room.
Ken Rice
krice at suspicious.org
Thu May 29 11:48:07 PDT 2008
With FreeSwitch there are a couple of ways to accomplish what you are doing
with 3 distinct levels of performance
Way 1) Full Media Interaction/Transcoding. This is very similar to the way
asterisk works and should on modern several give you atleast 2 to 3 times
the performance you see on asterisk (not accounting for any transcoding load
you may introduce)
Way 2) Media Proxy mode. In this mode you will see a good bit of performance
gain as FreeSwitch will only proxy the media it will not interact with the
media stream (ie: no transcoding, no DTMF events etc) but you can still cut
thru nat, appease providers that don't want to "hairpin" the media on their
networks, and still do a full topology hide (not applicable for your
scenario below as you can no jump in and out of proxy only mode)
Way 3) No Media Mode. In this mode FreeSwitch functions more along the lines
of openser/ser minus the media proxies. Media is passed directly between the
end points and FreeSwitch is completely out of the media path. This is the
most efficient mode for routing calls as there is no media load on
freeswitch and the number of concurrent calls is limited by system memory
resources and speed of the calls coming in (as in how many calls/second can
freeswitch process) for your particular application originating calls would
start in media mode and them move to no media mode once the second leg
starts to come online. This would have an impact of performance based on the
total number of calls doing media with freeswitch at any given time.
Now for some real numbers...
I route calls primarily using the no-media-mode using dell 1950s with Dual
QuadCore 2Ghz E5335's w/ 4Gs of ram. (admittedly this is a slightly
different method from what you are doing) in our configuration we are able
route in excess of 200 calls/sec with a concurrent call load in excess of
3000 calls (6000 legs) per machine. Where we run into problems is not in the
concurrent call volume, its in the Calls/Sec luckily FreeSwitch has a
Sessions/Second Limiter built in and we can set this and keep the box from
melting down.
Please Note in the above configuration we are largely routing autodialer
traffic so performace should be much better if you decrease the calls per
second and increase the average call length. High Call Per Second Rates are
the bane of any switch
K
> From: Nicolas Brenner <nicolas at medularis.com>
> Reply-To: <freeswitch-users at lists.freeswitch.org>
> Date: Thu, 29 May 2008 12:54:07 -0400
> To: <freeswitch-users at lists.freeswitch.org>
> Subject: Re: [Freeswitch-users] Max of 170 channels in the conference room.
>
> Hi, sorry for my ignorance, but I was wondering if these figures are
> in any way comparable to the performance FS would have doing bridged
> calls?
>
> I have a web callback app that's currently running on top of Asterisk,
> and I'm planning on moving to FS, and use originate/&bridge commands
> to bridge calls between two actual phones. I'd like to know if (using
> the same setup as Johny) I'd be able to hit more than 170 channels?
> (or more than 170 calls, I guess that would be 340 channels).
>
> Thanks
>
> On Thu, May 29, 2008 at 12:19 PM, Brian West <brian at freeswitch.org> wrote:
>> These aren't really dual core CPU's they are single core with
>> hyper-threading. If you disable hyper-threading you'll get more
>> performance.
>> You'll never get that great of performance out of these CPU's. The new
>> 64bit woodcrest/clovertown Xeon's are much better... night and day
>> difference. (Pinto vs Porsche)
>> /b
>> On May 29, 2008, at 11:10 AM, Johny Kadarisman wrote:
>>
>> Hi Brian, attached is my cpu info.
>>
>> Rgds,
>>
>> ================
>> processor : 0
>> vendor_id : GenuineIntel
>> cpu family : 15
>> model : 2
>> model name : Intel(R) Xeon(TM) CPU 2.80GHz
>> stepping : 9
>> cpu MHz : 2784.780
>> cache size : 512 KB
>> physical id : 0
>> siblings : 2
>> core id : 0
>> cpu cores : 1
>> fdiv_bug : no
>> hlt_bug : no
>> f00f_bug : no
>> coma_bug : no
>> fpu : yes
>> fpu_exception : yes
>> cpuid level : 2
>> wp : yes
>> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
>> cmov pat pse36 clflush dts acpi mmx fxs
>> r sse sse2 ss ht tm pbe cid xtpr
>> bogomips : 5573.58
>> clflush size : 64
>>
>>
>> processor : 1
>> vendor_id : GenuineIntel
>> cpu family : 15
>> model : 2
>> model name : Intel(R) Xeon(TM) CPU 2.80GHz
>> stepping : 9
>> cpu MHz : 2784.780
>> cache size : 512 KB
>> physical id : 0
>> siblings : 2
>> core id : 0
>> cpu cores : 1
>> fdiv_bug : no
>> hlt_bug : no
>> f00f_bug : no
>> coma_bug : no
>> fpu : yes
>> fpu_exception : yes
>> cpuid level : 2
>> wp : yes
>> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
>> cmov pat pse36 clflush dts acpi mmx fxs
>> r sse sse2 ss ht tm pbe cid xtpr
>> bogomips : 5569.09
>> clflush size : 64
>>
>>
>> processor : 2
>> vendor_id : GenuineIntel
>> cpu family : 15
>> model : 2
>> model name : Intel(R) Xeon(TM) CPU 2.80GHz
>> stepping : 9
>> cpu MHz : 2784.780
>> cache size : 512 KB
>> physical id : 3
>> siblings : 2
>> core id : 0
>> cpu cores : 1
>> fdiv_bug : no
>> hlt_bug : no
>> f00f_bug : no
>> coma_bug : no
>> fpu : yes
>> fpu_exception : yes
>> cpuid level : 2
>> wp : yes
>> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
>> cmov pat pse36 clflush dts acpi mmx fxs
>> r sse sse2 ss ht tm pbe cid xtpr
>> bogomips : 5569.31
>> clflush size : 64
>>
>>
>> processor : 3
>> vendor_id : GenuineIntel
>> cpu family : 15
>> model : 2
>> model name : Intel(R) Xeon(TM) CPU 2.80GHz
>> stepping : 9
>> cpu MHz : 2784.780
>> cache size : 512 KB
>> physical id : 3
>> siblings : 2
>> core id : 0
>> cpu cores : 1
>> fdiv_bug : no
>> hlt_bug : no
>> f00f_bug : no
>> coma_bug : no
>> fpu : yes
>> fpu_exception : yes
>> cpuid level : 2
>> wp : yes
>> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
>> cmov pat pse36 clflush dts acpi mmx fxs
>> r sse sse2 ss ht tm pbe cid xtpr
>> bogomips : 5569.67
>> clflush size : 64
>>
>>
>> On Thu, May 29, 2008 at 11:49 AM, Brian West <brian at freeswitch.org> wrote:
>>>
>>> Can you cat /proc/cpuinfo and post that?
>>> /b
>>> On May 29, 2008, at 10:41 AM, Johny Kadarisman wrote:
>>>
>>> Thanks Anthony,
>>>
>>> It's a 32bits Ubuntu server version, and I had blindly follow and run the
>>> test with following settings :)
>>>
>>> ulimit -c unlimited
>>> ulimit -d unlimited
>>> ulimit -f unlimited
>>> ulimit -i unlimited
>>> ulimit -n 999999
>>> ulimit -q unlimited
>>> ulimit -u unlimited
>>> ulimit -v unlimited
>>> ulimit -x unlimited
>>> ulimit -s 244
>>> ulimit -l unlimited
>>> ulimit -a
>>> bin/freeswitch
>>>
>>> also, I set with 120ms interval, I still hit around 170 channels limits in
>>> one room.
>>> Let me play around more with multiple room setup.
>>>
>>> Rgds,
>>> Johny K.
>>>
>>>
>>>
>>>
>>> On Thu, May 29, 2008 at 10:56 AM, Anthony Minessale
>>> <anthony.minessale at gmail.com> wrote:
>>>>
>>>> We don't do much testing on the conference but i can give you a few
>>>> pointers.
>>>>
>>>> is it a 32 or 64 bit box?
>>>> if it's 32 you can try this as root before you start up.
>>>>
>>>> ulimit -s 244
>>>>
>>>> you can also change the conference interval to higher number of ms
>>>> between packets to give it more time to mux the audio, the default is 20
>>>> but
>>>> you should be able to use 30,40,60,120 as well
>>>>
>>>> You can also try multiple conference to compare.
>>>>
>>>>
>>>>
>>>> On Thu, May 29, 2008 at 8:57 AM, Johny Kadarisman <jkr888 at gmail.com>
>>>> wrote:
>>>>>
>>>>> Hi all,
>>>>>
>>>>> I build a new test boxes for freeswitch. and trying to simulate high
>>>>> load condition.
>>>>> I have one freeswitch box that host conference apps, and another box
>>>>> with simple js to originate call to the conference room.
>>>>> To monitor the sound quality, I have one phones that dial into the
>>>>> conference room, and then start streaming/playing moh music in the
>>>>> conference.
>>>>>
>>>>> In many scenario that i tried (mute, unmute, one or multiple rooms), the
>>>>> max total number of channels is about 170 channels before FS start to
>>>>> slowing down on creating new channels. The Cpu's only tops around 40-50%,
>>>>> and the voice/rtp start to jitter or delays on my phone.
>>>>>
>>>>> These test boxes is running on 2 processor 2.8GHz Dual Core xeon cpu's
>>>>> and about 1Gb memory.
>>>>> I think the number already pretty impressive, at least for me :)
>>>>>
>>>>> But is this limit sounds right from others experience? Can I still do
>>>>> more tuning?
>>>>>
>>>>> Thanks,
>>>>> Johny K.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Anthony Minessale II
>>>>
>>>> FreeSWITCH http://www.freeswitch.org/
>>>> ClueCon http://www.cluecon.com/
>>>>
>>>> AIM: anthm
>>>> MSN:anthony_minessale at hotmail.com
>>>> GTALK/JABBER/PAYPAL:anthony.minessale at gmail.com
>>>> IRC: irc.freenode.net #freeswitch
>>>>
>>>> FreeSWITCH Developer Conference
>>>> sip:888 at conference.freeswitch.org
>>>> iax:guest at conference.freeswitch.org/888
>>>> googletalk:conf+888 at conference.freeswitch.org
>>>> pstn:213-799-1400
>>>> _______________________________________________
>>>> 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
>>>
>>> Brian West
>>> sip:brian at 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
>>>
>>
>> _______________________________________________
>> 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
>>
>> Brian West
>> sip:brian at 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
>>
>>
>
>
>
> --
> Nicolás Brenner
> Medularis SpA
>
> _______________________________________________
> 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
More information about the FreeSWITCH-users
mailing list