[asterisk-users] Asterisk concurrent calls count

[Steve Totaro]

On Sun, May 18, 2008 at 1:24 PM, Tilghman Lesher
<tilghman at mail.jeffandtilghman.com> wrote:
> On Sunday 18 May 2008 10:56:00 Jay R. Ashworth wrote:
>> On Sun, May 18, 2008 at 02:45:09AM -0400, Steve Totaro wrote:
>> > I do however, think that Digium should provide some rough concurrent
>> > call figures and I guess that is how I got off topic on this SwitchVox
>> > tangent.  There are some common feature sets, especially when looking
>> > at PBX functionality with or without Zap or transcoding hardware that
>> > could be published (with a disclaimer of course).  There are also
>> > common server platforms but that is more of a moving target.
>>
>> Sure.  But can you understand the point Tilghman was making there?
>> They quote those numbers as "cover our ass" numbers; their intention is
>> to avoid running up their tech support expense too much.
>
> An additional point worth making is that the numbers for Business Edition
> aren't based on installation on a very specific machine, where we've tweaked
> operating system and Asterisk variables for optimal performance, for the
> express purpose of coming out on top of stated benchmarks.  They are call
> numbers that you can expect will be stable, even if you aren't good at finely
> tweaking Linux systems, as, I suspect, most people running production systems
> are not.
>
> I can't say that Digium will never publish benchmarks, but I believe that most
> people who are concerned with the proper functioning of whatever they've come
> to use Asterisk for (call routing, PBX, IVR, or something else) take
> benchmarks with an extremely large chunk of salt.
>
> --
> Tilghman
>

Asterisk benchmarking is one of the topics that comes up on the list
frequently and consistently for what, like the last six years (that I
have been involved with Asterisk)?  I would call that a "Salt
Boulder".

On the wiki there is a long page dedicated to dimensioning.  I know
beyond a shadow of a doubt that people really are very interested in
the numbers.

If Digium won't supply benchmarking, then let's have a 3rd party throw
down the gauntlet.

Digium vs Sangoma on stock kernels and stock machines with identical
call load, apps, transcoding.  No optimization, just stock, and see at
what load we hit a breaking point.

Then Asterisk vs FreeSwitch on similar apps pushing them to the
breaking point as well.

Nothing but pure numbers.

Thanks,
Steve Totaro