asterisk users - Jul 2004 - AW: Large Enterprises using asterisk

Hi,
> I've never run against a commercial PBX that didn't need 
> maintenance.
Acknowledged.
>  VM hard 
> drives fail,
> ...
> Asterisk is 
> every bit as stable 
> as the old-gen KSUs and PBXs.
There are big differences. As I know of no other PBX that uses
'consumer' hardware, asterisk is also struggling with problems in the
underlying Hardware. And even Software. The NEC/Nitsuko PBX I bought for our
company 4 years ago has still no problems beeing extended. Try this with a four
year old version of Linux underneath your asterisk.

Don't get me wrong, ist also a BIG advantage of the * solution to always be
able to upgrade to HEAD without 'licensing' upgrades. Also the hardware
upgrades to a decent PC platform is a LOT cheaper than any hardware upgrades
commercial PBXs offer.

BUT.

What are the key points if asterisk wants to make it into many offices and
companies? I sorted them in the order they mattered to me:
1. Reliability. Triple nine is not enough as for almost every company, phones
are the vital wire to the outside world. Here asterisks suffers major
disadvantages to 'closed' solutions commercial PBXs offer. This is due
to its complexity, due to its speed of development and change and due to its
underlying Hard- and Software.
2. Reliability. ;-) a second aspect of reliability goes further than the
'technical' reliability I already mentioned. If a CIO in a company
decides that asterisk is beeing rolled out as a telephony solution, then he has
to be absolutely convinced, that the project and the development of asterisk is
guaranteed for multiple years (continuity). Or else he buys Cisco or IBM, nobody
is fired because of buying Cisco or IBM, you know this topic.
3. Extendability. Here is big point number one for asterisk. No PBX I know has
even such an impressing static list of features. So nothing to say about the
dynamic changes in evolving features.
5. Ease-of-use. Asterisk with its configuration files has managed to stay
manageable despite of the huge feature list. So no problem from operator side.
For the consumer this is quite different, because system phones have a whole lot
of functions (and function keys etc.) that a 'vanilla' phone isn't
able to manage, even as asterisk provides all of that functions. So here the
proprietary vendors score a point.  Innovaphone for example has the solution
that ONE hardphone is guaranteed to work perfectly with their VOIP-PBX since the
protocol extensions it uses are supported, * should think about something alike
...
4. TCO. Asterisk with the underlying structure of PC hardware and open source
operating system has a big advantage here. Also system and
network-administrators with Linux or BSD knowhow are a common must-have for many
companies so we don't suffer any disadvantage here.

I am CIO of a small company (~80 phones) where the actual PBX has reached its
limits. We decided against Asterisk because of the two 'reliability'
points. PC Hardware fails. Asterisk has Bugs (I had several crashes with
voicemail functions, perhaps codec-related as it only happened with some
VOIP-Softphones) and Asterisk and Linux evolve too quick to follow without
remarkable manpower.
I would very much like to rethink our decision next year, but at the moment, I
would not rollout a vital infrastructure based on asterisk (we will continue our
asterisk evaluation, but only fore some teleworkers).

GNU/Linux has the same problem in many companies. The big distributors as Red
Hat or SuSE address this by guaranteeing continuity in there 'enterprise
server' products, Debian does the same with the 'stable' branch.

So my proposal is, that with the 1.0 version let a distribution form AND an
attitude of 'stability and continuity' in that branch do the job of
convincing managers (not more than one major release in two years). And let an
evolving base of features in the HEAD branch convince the technicians. But this
means a LOT of back- and forward porting of patches and managing of developers,
who tend to only fight on the 'bleeding edge', especially when not being
paid for their work. So this is something only Mark or Digium could achieve or a
person/team that has full support by them.


Just my 2 Euro-Cents, and sorry for the long text, hope somebody made it to the
end ... ;-)

Stefan M?rkle

-- 
Stefan M?rkle                       Netpioneer GmbH
Leitender Systemarchitekt           Beiertheimer Allee 18
<stefan.maerkle@netpioneer.de>      76137 Karlsruhe