[Asterisk-Users] Intel Modem vs Digium Cards

[Joe Greco]

> > Wow, that's a really sucky attitude.  I would expect *Digium* to tell him
> > to go away and solve his own problems.  However, if the user community does
> > that, then this is one of the suckiest user communities I've run across in 
> > the free software world, and I've been doing free software for many years.
> 
> Well, IMHO, I would expect it perfectly reasonable for one of three
> responses from the 'user' community (in order of likelihood):
> 
> 1) A resounding non-response
> 2) A response of "Well, get a X101P or TDMx0P and try it
> 3) Some advice on how to resolve the problem.

All of those are quite reasonable, of course.

The "bad attitude" response to that user, however, seemed quite uncalled-
for.

> Reasoning:
> Most people don't use non digium hardware, so they have no experience
> with the intricacies that might go wrong for you, whether it is a
> hardware problem, driver incompatibility, or whatever, we don't know the
> hardware, we don't trust the reliability of the solution, etc... This is
> why you would most likely get one of the first two responses.
> 
> Sure, there are some people who have used the hardware on this list,
> there are some people who haven't but will attempt to help anyway by
> presuming that your hardware/driver is all fine, etc...There may even be
> some people who know enough to tell you that the hardware you have won't
> work, or that you need to apply some obscure patch...
> 
> Finally, when you are advising a NEW user, to add additional
> complications/limitations as the above, you are asking for trouble.
> Sure, some people could express their opinions
> better/differently/whatever, but at the end of the day, you have a few
> potential outcomes:
> 
> The user buys a generic modem
> a) can't get asterisk to work after some days/weeks, and gives up, tells
> everyone that asterisk is crap, immature, and doesn't work

Yes, that seems bad.  However, free software projects always have that
problem.  Here's what really seems to happen:

User X has <whatever> experience with Package P.  User X says <anything>
about Package P to User Y, who has never heard of Package P.  The general
outcomes of that are not particularly negative - if Y has not heard of P,
that's bad.  If X conveys a positive experience to P, that's good.  If X
conveys a negative experience to P, well, you're not really any worse off
than when Y hadn't heard of P, but in fact Y may then still look at P
(and even come to a different conclusion).

The old saying about "print whatever you like about me, just be sure to
spell my name right" applies to free software projects quite well.

> b) gets asterisk to work, but has major stability problems, or even just
> minor annoying hangups/echo/whatever and so they drop it and tell people
> that it's crap

See above.  I'll further note that this sort of thing typically plagues
free software projects, and that people expect them to eventually be
worked out.  It won't stop many people from investigating.

Further, the X100P is itself apparently mainly a mild variation on a
standard chipset design.  That leaves some doubt as to whether or not
the experience would be significantly different between the X100P and
the generic.

> c) gets asterisk to work, and everything runs really smoothly. They tell
> everyone asterisk is great, and there is no need to buy the $100 card,
> then we propogate all of the above chances for failure.... 

Except that neither the $100 card nor the $5 card are viable in large
quantities in a single system, according to recent posts to this very 
list.

The guy who's been tasked by his boss to look at Asterisk is likely to
buy the more expensive card, /because/ of the support.

The guy who's playing with it at home is more likely to buy the $5 card,
and may have troubles, but at least he'll be playing with Asterisk,
rather than not playing with it.

In both cases, what I've read leads me to believe that it won't be
feasible to have large numbers (specifically, many more than two) of
the X100P's in a single system.

At that point, then, you're largely looking at people buying the X100P
as a platform for experimentation.  They'll either be moving on to some
TDM400P's or a T100P with a channel bank (etc) in order to implement 
that production system.

So option c) doesn't really concern me /too/ much.

I personally believe that Digium chose to implement things this way so
that people would have some options.

> > As for me?  I'm shopping for cheap modem cards.  Why?
> > 
> > 1) I'm on FreeBSD, so Digium probably won't support that.
> 
> (I'm not saying freebsd is bad/better/worse/whatever).
> So, you are limiting the number of people who can assist you to those
> who have used FreeBSD, and then again to only those who have used a
> generic modem.....

I'm even further limiting myself because I really don't care too much if
I manage to get it to work, because I probably won't put that much effort
into it, and in fact may not even buy it to begin with, so that's an even
bigger limit.  :-)

> BTW, I though digium was actually putting a reasonable amount of effort
> into getting asterisk + drivers to work on FreeBSD, and have seen a
> number of people claiming success.

I was under the impression that the Zap driver port was being undertaken
independently.  I may be mistaken, as I haven't been closely following 
it.  

> > 2) I realistically expect to go all VoIP, except perhaps for fax, so I don't
> >    want to spend a ton on cards that I won't need.
> 
> If you need them, buy them, if you don't, don't. I don't see why you
> would buy something that you 'might' need, regardless of the price??

We're a technology company.  Experience is useful.  Ten dollars would not
be our largest speculative purchase this year, not by a long shot.

> > 4) I don't really think my PPro200 PBX box will survive very well with
> >    having to handle the codec work anyways.
> 
> There are other people that have had success with that sort of machine,
> depending on the number of channels/codec conversions etc...

Right.  :-)

> > But I'm open to spending ten bucks to explore this method. 
> > 
> > If I was buying a Digium card and it didn't pan out, I'd probably want to
> > see if I could return it, and then there's all the annoyance of an RMA, and
> > time frames after which you can't return it, etc.  This way, I'm out a
> > whopping $10.90, and I can deal with that.
> 
> Well, I would expect that you wouldn't need to RMA it, digium seem
> pretty good at supporting their products... but, I don't want to spoil
> your fun, get the cheap modem, worst case is you will either buy the
> X101P, or will go away and tell people how bad asterisk is....

Well, that's unlikely, as I'm not convinced Asterisk is bad.  As a matter
of fact, it's probably the least problematic bit of VoIP technology I've
met up with.  X-Lite used to blow up Windows PC's after several restarts
(and the current version still lags the hell out of them).  The Cisco
7960 and the licensing hell that Cisco seems intent on burdening everyone
with definitely counts as no fun.  The Sipura 3000 worked great until I
firmware upgraded it and it went south for the winter.  Our VoIP provider
misconfigured its end of things, causing much grief.  Asterisk has had
its quirks, but most are documented, and the remainder just require a bit
of experimentation and lots of reading.

> > Oh.  That's over at ChiefValue.com.  Encore 56K V.92 Internal PCI Fax Modem,
> > Model ENF656-ESW-INPR - Retail.  $5.90 plus $5.00 shipping.
> 
> I wish you luck, I know nothing about these generic modems, but it
> doesn't look like that even comes close to mentioning the right
> indicators to show that it would work.... but, like I said, I don't
> know.

http://www.voip-info.org/wiki-Asterisk+Hardware

says:  Intel modem marked with chipset 537 or MD3200. This card installs and
behaves exactly like a Digium Wildcard X101P card.

http://www.encore-usa.com/product.php?id=38&lang=

says: The ENF656-ESW-INPR is a V.92 modem chipset that is built with Intel
537 Chipset, optimized for PCI bus specification. 

What indicators is it missing?  I spent a fair amount of time digging
around, looking specifically for modems with the 537 or MD3200.

> > Share the knowledge.  It's not bad for Digium.  The guy who wants to get
> > into this telephony stuff for a cheap price, or just wants to see if it
> > will work, he's going to be attracted by the ten dollar deal.  Let him do
> > that and then preach the glories of Asterisk.
> 
> Or preach how crap asterisk is....

"Just spell my name right"

> > Remember this:  Digium can't grow (much) unless the community grows.  So
> > help it grow.
> 
> By advocating sustainable growth methods... ie, promoting known good
> hardware, which can be easily supported. Better options that the X101P
> clones exist, such as the various ATA devices, or for ISDN there are
> even more options. Look on the wiki for more, there is a lot of
> information available there.

Well, I'm probably biased towards VoIP, so I'd even agree that better 
options than even the X100P itself exist.

However, I'm curious what better options there are for ISDN.  If you're
talking PRI, yeah, that's nice, Digium has a good solution.  However, for
those of us here in the US, we're burdened with having to find NI1
compatible resources, and I've not seen anything that I'm convinced will
work with Asterisk.

At this point, it looks like the most hopeful way to directly connect a
BRI via VoIP would be a Cisco VIC2-2BRI-NT/TE in an appropriate chassis.

What I'd really like is some way to be able to terminate some V.90 calls,
a neat capability we currently have on BRI by using a USR I-Modem in
V.90 server mode.  The information about doing stuff like this is pretty
sketchy to begin with.  I have a fair amount of flexibility.  If possible,
I would handle it as a BRI in from the LEC, into the Cisco, through 
Asterisk, and back out the Cisco to the I-Modem on a network-configured 
port.  Seems like a good theory.

> Just my 0.02c worth.
> 
> PS, for those people who feel like sending a message to the list saying
> "Don't use clone x101Ps, or whatever, please don't. The best response
> you can send is a non-reply. Those people who want to support that user,
> will, the rest of us should just keep quite. The same as you would keep
> quite if you know nothing about a sipura and you see a post about some
> wierd problem someone is having with their new sipura.

In other words, the same common courtesy generally expected on other lists.

> PPS, thanks for reading my drivel :)

Heh.

Regards,

... JG
-- 
Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net
"We call it the 'one bite at the apple' rule. Give me one chance [and] then I
won't contact you again." - Direct Marketing Ass'n position on e-mail spam(CNN)
With 24 million small businesses in the US alone, that's way too many apples.