[asterisk-users] Asterisk concurrent calls count

[Tilghman Lesher]

On Friday 16 May 2008 09:11:11 Steve Totaro wrote:
> On Fri, May 16, 2008 at 9:56 AM, Tilghman Lesher wrote:
> > On Friday 16 May 2008 06:59:15 Al Baker wrote:
> >> this is one very weak area for *. There is NO ANSWER.
> >> Now in fairness to *, the answer DOES depend on a # of critical
> >> variables. How much CODEC to CODEC transcription is going on.
> >> How many MEET Me conferences are going on.
> >>
> >> On the other hand, DIGIUM COULD, since they have a lab take 4-5
> >> 'standard' workloads
> >> on two of the most common hardware boxes, say Dell & HP, and run x # of
> >> transcriptions and
> >> show the #'s.
> >> Then x # of meet-me conferences.
> >>
> >> Face it the DB Industry did this 15-2- YEARS ago with TP benckmarks
> >>
> >> Rockwell and NORTEL can tell you this for every piece of hardware they
> >> sell.
> >>
> >> It is a an area DIGIUM need to "man-up" in.
> >
> > I'm not sure what your problem is with Digium.  They sell several
> > machines for which they publish very specific numbers as to how many
> > users those machines will support (the Switchvox appliances).  Note that
> > these machines are configurable only from the web interface, and they do
> > not allow you to install additional software.  In other words, when they
> > give you a specific machine, with a ton of those variables controlled,
> > they can give you a number.
> >
> > Digium is under no obligation to give you numbers for your own hardware.
> > That's up to you (and you get to control your own set of variables).
>
> It seems any constructive criticism offered, you take as an attack
> against Digium.  That is not a good attitude.

I don't see how you figured out what I was thinking.  Al said Digium doesn't
publish any numbers, and I responded, saying that he was incorrect; Digium
does indeed publish numbers (they're just not for his hardware).

> While under no obligation, it certainly would help sales.

Whose sales?  If you're talking about the appliances, then yes, I'm sure the
publication of those numbers help with sales.  If you mean your own sales,
well, you're right, Digium's numbers probably don't help your sales.  You
could certainly put together a lab and do your own testing.  Why don't you do
that?

> I take "Appliance Numbers" with a grain of salt.  The sales model of
> SwitchVox (and most others) is based on number of ports (SoHO, SMB,
> Enterprise) not maximum number of ports that the appliance could
> actually handle if not artificially constrained.

Consider the maximum number of ports that Switchvox will enable on a single
machine and consider that the maximum number that they're willing to support
comfortably without running into some hard limit.  You never want to run into
a hard limit in the field anyway.

> This is in the style of legacy proprietary systems and anther reason
> why the sale cycle goes a little tougher than a custom job.  Asterisk
> with FreePBX (and maybe Druid) eliminate these artificial constraints
> on usage.

Yes, but the point of those constraints is to permit support a manageable
job.  Yes, you could probably add 2 or 3 or 10 or 15 to the number of calls
that a particular machine could handle, but from a support perspective, it
doesn't matter how many the machine could theoretically handle, it matters
how many it could handle in the particular installation in a supportable
configuration (those are all those pesky variables we've been talking about).

> I have load averages and CPU usage stats in my mind for all the
> various usages and hardware through experience in my mind.  Of course
> they are only valuable to the exact setup I was doing.

Precisely.

-- 
Tilghman