[asterisk-users] Ast/Hyla/IAX Scalability?

[Jeff LaCoursiere]

On Fri, 13 Mar 2009, David Backeberg wrote:

[various snippage]

> Again, you'll find people arguing that their voip solution has as low
> of a failure rate as a hardware solution. I'm jealous. My voip fax
> solution does not yet have that low of a failure rate, but I'm
> hopefully getting closer to working out the last bugs.

The proposed platform being enclosed in one machine should make it as good 
as a hardware solution IMO.  Certainly cheaper.  Don't have to worry about 
a board frying and taking the whole service down.

>
>> I've also noticed that IAXmodem is compiled statically against a version of
>> spandsp included with the iaxmodem source. For a large installation, would
>> it be better to compile iaxmodem dynamically to reduce the per-instance size
>> of each iaxmodem? Or, would it be better to simply throw more RAM at it?
>
> I'm not sure what difference RAM makes. What breaks a fax on voip is
> latency and dropped packets.

There won't be any latency or dropped packets because there are no 
physical network links involved, though if the machine becomes too loaded 
or starts swapping for lack of RAM, more RAM would make all the 
difference.  RAM is so cheap anyway you may as well load it up!  I don't 
think compiling dynamically would save enough RAM to make a difference 
anyway... a few MB?  My iaxmodems seem to reserve 3MB each, and their 
binary size is only .5MB, so some piece of that .5MB could be saved per 
instance.  The upper limit saved over 45 instances would be something less 
than 23MB then... Someone will probably point out the error in my 
calculations, though :)

>> Are there any concurrency issues when receiving a large number of faxes on a
>> system using IAXmodems?
>
> File system contention, but fax files aren't very large, and I would
> call that a non-issue. Most people don't want a piece of paper; they
> want a PDF that they can turn into paper once in a while.

File system contention won't break faxes, though, as at that point the fax 
is already received.  I think at some point CPU contention will start to 
appear, as each instance is doing some hefty (?) DSP work.  Just guessing 
though.  I run this same setup, but have never seen more than four or five 
at once.  Damn stable and error free though!  I lose about one page in a 
thousand, and that seems to always be the same senders, so I suspect a 
prob on their end.

j