[asterisk-biz] Using SUN T1000 or T2000

Sun Jun 4 12:51:03 MST 2006

Hi there,

Using SUN T1000 or T2000 would not make sense for Asterisk. These are
awesome machines for highly multithreaded applications like web servers or
J2EE. But the CPU is sharing with it's cores only one floating point unit.
If you are doing all your codec translations it would be pointless to have
an 8 core system with one floating point unit. The performance might be
worse than a regular AMD or Intel single core chip (no idea). What would be
interesting is the performance of running Asterisk on UltraSparc IV+. I bet
that would be really fast especially since you can scale it almost linear.
Beyond the bottom line you will get the best results at the best
cost/performance with AMD boxes (from SUN if you will;-) like x2100) and
scale horizontal (more boxes for more calls). And a hardware g729 board from
Digium would solve all these problems completely. With a nice DSP hardware
based solution you can probably handle hundreds of calls on one board
without the PC CPU doing anything. Can't wait. Maybe Cisco is breaking out
in sweat then...That's whats making their 5400 servers a good choice for SIP
termination to put 700 calls in 2U. Two AMD servers are way cheaper and
consume probably less power as well (almost the highest portion of our cost
by the way). 

Best

Nikolai
CEO, nikotel Inc.

On 6/4/06 11:02 AM, "asterisk-biz-request at lists.digium.com"
<asterisk-biz-request at lists.digium.com> wrote:

> Send asterisk-biz mailing list submissions to
> asterisk-biz at lists.digium.com
> 
> To subscribe or unsubscribe via the World Wide Web, visit
> http://lists.digium.com/mailman/listinfo/asterisk-biz
> or, via email, send a message with subject or body 'help' to
> asterisk-biz-request at lists.digium.com
> 
> You can reach the person managing the list at
> asterisk-biz-owner at lists.digium.com
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of asterisk-biz digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: ASTCC Developer (alex at pilosoft.com)
>    2. Re: Performance on Pentium Dual Core CPUs (Sergey Kuznetsov)
>    3. Re: ASTCC Developer (Bruce Komito)
>    4. RE: Performance on Pentium Dual Core CPUs (Boris Bakchiev)
>    5. Re: Performance on Pentium Dual Core CPUs (Kevin P. Fleming)
>    6. RE: Performance on Pentium Dual Core CPUs (Boris Bakchiev)
>    7. Re: Performance on Pentium Dual Core CPUs (Sergey Kuznetsov)
>    8. Re: Performance on Pentium Dual Core CPUs (Tomer Horn)
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Sun, 4 Jun 2006 10:37:26 -0400 (EDT)
> From: alex at pilosoft.com
> Subject: Re: [asterisk-biz] ASTCC Developer
> To: Commercial and Business-Oriented Asterisk Discussion
> <asterisk-biz at lists.digium.com>
> Message-ID:
> <Pine.LNX.4.44.0606041037040.29748-100000 at bawx.pilosoft.com>
> Content-Type: TEXT/PLAIN; charset=US-ASCII
> 
> On Sun, 4 Jun 2006, Sahil Gupta wrote:
> 
>> I need a few things modified on the current version of astcc.  If there
>> is someone competent, please contact me off-list.
> Sahil,
> 
> Can you reply to my emails - not sure if you are receiving them.
> 
> Thanks.
> 
> 
> 
> ------------------------------
> 
> Message: 2
> Date: Sun, 04 Jun 2006 10:50:52 -0400
> From: Sergey Kuznetsov <asterisk_biz at deeptown.org>
> Subject: Re: [asterisk-biz] Performance on Pentium Dual Core CPUs
> To: Commercial and Business-Oriented Asterisk Discussion
> <asterisk-biz at lists.digium.com>
> Message-ID: <4482F34C.7080504 at deeptown.org>
> Content-Type: text/plain; charset=UTF-8; format=flowed
> 
> Kevin P. Fleming wrote:
>> ----- Craig Lawrence <craig at mytel.net.au> wrote:
>> 
>>   
>>> PS. Dual processors are overkill unless you have an Asterisk version
>>> that supports it (we don't).
>>>     
>> 
>> I don't know what you are talking about here... Asterisk is a heavily
>> threaded application, and will take full advantage of the number of
>> processors/cores in your system. Especially when transcoding is involved,
>> having multiple cores available makes a significant difference.
>> 
>> This has been true for all released versions of Asterisk; there was never a
>> released version of Asterisk that didn't 'support' multiple processors.
>> 
>>   
> Kevin,
> 
> Can you share your expectations how many G729 transcodings can be done
> on dual dual-core Opterons or P4 with 4 Gb of memory on SIP-to-SIP
> environments ( no TDM or echocancelers involved)
> My guess is about 250-300 simultaneous calls.
> 
> May be I am way to optimistic, that's why I am asking your educated
> expectations.
> 
> 
> All the Best!
> Sergey.
> 
> 
> 
> ------------------------------
> 
> Message: 3
> Date: Sun, 4 Jun 2006 08:17:47 -0700 (PDT)
> From: Bruce Komito <brucek at bagel.com>
> Subject: Re: [asterisk-biz] ASTCC Developer
> To: Commercial and Business-Oriented Asterisk Discussion
> <asterisk-biz at lists.digium.com>
> Message-ID: <20060604081708.T75213-100000 at mustang.bagel.com>
> Content-Type: TEXT/PLAIN; charset=US-ASCII
> 
> Sahil, We have some experience in that area.  Please contact me if you
> wish to discuss further.
> 
> Regards
> 
> Bruce Komito
> High Sierra Networks, Inc.
> www.servers-r-us.com
> (775) 236-5815
> 
> 
> On Sun, 4 Jun 2006 alex at pilosoft.com wrote:
> 
>> On Sun, 4 Jun 2006, Sahil Gupta wrote:
>> 
>>> I need a few things modified on the current version of astcc.  If there
>>> is someone competent, please contact me off-list.
>> Sahil,
>> 
>> Can you reply to my emails - not sure if you are receiving them.
>> 
>> Thanks.
>> 
>> _______________________________________________
>> --Bandwidth and Colocation provided by Easynews.com --
>> 
>> asterisk-biz mailing list
>> To UNSUBSCRIBE or update options visit:
>>    http://lists.digium.com/mailman/listinfo/asterisk-biz
>> 
>> This message has been categorized as "Legitimate" by Bayesian Analyzer.
>> If you do not agree, please click on the link below to train the Analyzer.
>> http://nospam.wpti.net/bt/a.aspx?M=C:%5Csmtpmail%5CBayesTraining%5C2006-06-04
>> %5C55f7593b76684e3f99152900aead3aa6&C=2
>> 
>> --
>> -----------------------------------------------------------------------
>> This message has been inspected by DynaComm i:mail
>> -----------------------------------------------------------------------
>> 
>> 
> 
> 
> 
> ------------------------------
> 
> Message: 4
> Date: Mon, 5 Jun 2006 02:52:58 +1000
> From: "Boris Bakchiev" <boris at jildent.com.au>
> Subject: RE: [asterisk-biz] Performance on Pentium Dual Core CPUs
> To: "Commercial and Business-Oriented Asterisk Discussion"
> <asterisk-biz at lists.digium.com>
> Message-ID:
> <13404687B15D66459DAC76C2AEFDE64D3A437B at mail1.jildent.com.au>
> Content-Type: text/plain; charset="us-ascii"
> 
> I'm surprised no one tested that for themselves yet.
> Mind you the test I did was with Intel based codec because I didn't want
> to buy so many codecs from Digium just to test the performance.
> 
> I did the test from g729 to alaw peers with round trip audio so both
> encoder and decoder was working hard.
> 
> In this scenario, correct me if I'm wrong, each call had following path:
> G729 client -> [asterisk transcode to slinear then to alaw] -> alaw
> client -> [asterisk transcode to slinear then to g729] -> G729 client.
> 
> I verified the path in sip show channels.
> 
> With 50 Calls CPU usage was:
> Cpu0  : 28.9% us,  7.5% sy,  0.0% ni, 57.2% id,  0.0% wa,  0.0% hi,
> 6.5% si
> Cpu1  :  7.0% us,  1.5% sy,  0.0% ni, 91.0% id,  0.0% wa,  0.0% hi,
> 0.5% si
> 
> I could not test any more concurrent calls, as my desktop for alaw peers
> was not fast enough.
> 
> As you can see, there is plenty of power in Pentium D and I presume that
> AMD CPU's would be even better since the bus between CPU's is not shared
> like in intel platform.
> 
> The system was Pentium D 830 coupled with 1GB PC2-6400 in dual channel
> mode (2x DDR2-800 memory sticks) on Asus P5WD2 Premium motherboard
> running software raid on SATA2 drives.
> 
> We noticed a big difference in system performance between PC2-5300 and
> PC2-6400 so if you can afford it, get PC2-6400.
> 
> Regards
> 
> -----Original Message-----
> From: asterisk-biz-bounces at lists.digium.com
> [mailto:asterisk-biz-bounces at lists.digium.com] On Behalf Of Sergey
> Kuznetsov
> Sent: Monday, 5 June 2006 00:51
> To: Commercial and Business-Oriented Asterisk Discussion
> Subject: Re: [asterisk-biz] Performance on Pentium Dual Core CPUs
> 
> Kevin P. Fleming wrote:
>> ----- Craig Lawrence <craig at mytel.net.au> wrote:
>> 
>>   
>>> PS. Dual processors are overkill unless you have an Asterisk version
>>> that supports it (we don't).
>>>     
>> 
>> I don't know what you are talking about here... Asterisk is a heavily
> threaded application, and will take full advantage of the number of
> processors/cores in your system. Especially when transcoding is
> involved, having multiple cores available makes a significant
> difference.
>> 
>> This has been true for all released versions of Asterisk; there was
> never a released version of Asterisk that didn't 'support' multiple
> processors.
>> 
>>   
> Kevin,
> 
> Can you share your expectations how many G729 transcodings can be done
> on dual dual-core Opterons or P4 with 4 Gb of memory on SIP-to-SIP
> environments ( no TDM or echocancelers involved)
> My guess is about 250-300 simultaneous calls.
> 
> May be I am way to optimistic, that's why I am asking your educated
> expectations.
> 
> 
> 
> ------------------------------
> 
> Message: 5
> Date: Sun, 4 Jun 2006 11:52:46 -0500 (CDT)
> From: "Kevin P. Fleming" <kpfleming at digium.com>
> Subject: Re: [asterisk-biz] Performance on Pentium Dual Core CPUs
> To: Commercial and Business-Oriented Asterisk Discussion
> <asterisk-biz at lists.digium.com>
> Message-ID:
> <17579559.13621149439966809.JavaMail.root at jupiler.digium.com>
> Content-Type: text/plain; charset=utf-8
> 
> 
> ----- Sergey Kuznetsov <asterisk_biz at deeptown.org> wrote:
> 
>> Can you share your expectations how many G729 transcodings can be done
>> 
>> on dual dual-core Opterons or P4 with 4 Gb of memory on SIP-to-SIP
>> environments ( no TDM or echocancelers involved)
>> My guess is about 250-300 simultaneous calls.
> 
> I don't have any first-hand experience, as we don't do performance
> benchmarking like that.
> 
> In any case, the amount of memory will make no difference at all. The clock
> speed of the CPUs and the speed of the memory bus will be the determining
> factors, so realistically I would expect the Opterons to perform better than
> P4s due to their much more efficient memory subsystem.
> 
> However, I can say that I doubt any existing PC-type platform will be able to
> achieve 250 or 300 simultaneous G.729 transcodes; any single box that could
> handle that would cost far more than a pair of lower-performance systems to do
> the same thing (think scaling horizontally, not vertically). In addition, when
> Digium releases the hardware G.729 transcoder board in the near future this
> will become less of an issue for deployments of that size.