[asterisk-users] Is Asterisk ready for Prime-Time?

[Norman Franke]

On Mar 19, 2008, at 1:00 PM, asterisk-users-request at lists.digium.com  
wrote:

>    Am I expecting too much?


Perhaps.

I think the hardware on which we run Asterisk can be much more  
reliable than the software, which is often the case. We have a bunch  
of HP servers with RAID and have never lost anything. A HD may fail,  
but the RAID keeps it going until we pop a new drive in there. A  
server class PC with redundant power supplies and RAID is really quit  
inexpensive now. If you are running on a $1000 box, you can't expect  
the reliability of dedicated telco hardware.

As for Asterisk, reliability has been a concern. Concurrency issues  
keep cropping up (read bugs.digium.com), especially with the SIP  
stack. This is particularly the case with buggy clients (soft phones,  
and under high volume of calls.) However, in fairness, writing  
heavily threaded code in C is very hard to get right. I think testing  
could surely be better, perhaps come code reviews and more guidelines  
for writing threaded code.

We had an old hardware system and it wasn't without some issues. We  
needed to support around 30 call takers and another 50 hard phones.  
It took us a while in the 90s to get everything working acceptably.  
Our transition time with Asterisk has actually been shorter. Since we  
have a highly customized operation, going with a Avaya or Cisco  
solution would have cost in excess of $500K. With Asterisk, we spent  
maybe $50K on hardware (including a Cisco gateway, two Asterisk  
servers and some Polycom phones.) This cost is trivial compared to  
how much we pend on our yearly phone bill.

The great benefit to Asterisk for us was that everything is open  
source software and thus we can customize it. We wrote a custom app  
that plugs into Asterisk that handles all of our custom business  
rules and provides far more capabilities than our old (and very  
expensive) hardware solution. Since we already had a custom developed  
desktop application, we could plug in a SIP stack and further  
customize things to be just what we wanted.

I remember talking to a rep from a large reseller and listing our  
requirements, and he was amazed we could do all we were going on 90s  
technologies, since their new (and even more expensive) stuff  
couldn't without lots of "consulting". We had just two developers  
over 6 months go from zero to a full call center solution.

On the other hand, if I were to support a small office with 20 people  
and simple voice mail for mission-critical telecommunications, I'd  
likely get a hardware solution. They are reliable and not that  
expensive. Asterisk, for now, and in my opinion, is always going to  
require more interaction that other hardware solutions. But, it's  
cheaper and more flexible. You may not care about cheap and flexible,  
and if not, maybe it's not what you want.

I've not tested products like CallWeaver or others. People claim some  
of these are more reliable, but Asterisk seems more popular.

Norman Franke
Answering Service for Directors, Inc.
www.myasd.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20080319/95288680/attachment.htm