[Asterisk-Dev] best approach for a new isdn driver mainly for asterisk.. ?!

[Andrea Venturi]

Steven Critchfield ha scritto:
> Quick question, Have you had good experience with USB devices under
> linux that push fair amounts of data? I ask this since I haven't. This
> list also shows that the USB stuff Mark has already been working with
> seems to be finicky about which chipset is under it. I suggest you
> double think the use of USB as a server level technology.
> 
> Don't take what I said before as a dismissal of your ideas, just a worry
> that you would end up with a lot of time invested in a product that may
> not be very stable due to parts outside of your project.
> 
> I do like the idea of seeing if the current ISDN adapters could be made
> to be more of a raw device and then link the libpri functions in a
> modified form for BRI to those devices. At that point a zap like
> interface to ISDN hardware should be easy. Who kows, maybe we could get
> to the point of supporting ISDN handsets which should be better than
> ADSI. 
> 

i have been waiting an answer from mark who, for sure, knows which way 
could be more feasible and smart, but i believe he's a little upset WRT 
all the potential "damage" if digium lose the HW revenue stream.

so i'd like to firmly state that:

- the price/performance ratio of the digium E1 cards is "unbeatable"!
- digium support is top level
- we already bought an E400P and an E100P and we are really amazed of 
the quality (did you know of the easter eggs under the chips on the board!?)
- we plan to buy more from them if we have to scale our main "pop" for 
IVR services

but we need to:

- deploy lighter IVR POPs with less channels than an E1 because we want 
leverage telco "cheaper" toll rates if we match the inbound prefix of 
our calling customers (in italy there's really a big fragmentation of 
telco zones and rates)
- stick with ISDN BRI because we don't trust POTS as a reliable telco 
link and it's too poor WRT features
- stick with USB ports because we would like to deploy diskless asterisk 
(based on a flavour of the epia via mainboard without PCI slots ) and 
USB is the only "theoric" choice for isochronous channel to ISDN (but it 
seems quite a dream or a nightmare)

so if digium would sell USB isdn adapters supported by *, we would be 
really glad to buy from them (not that large figures, anyway), otherwise 
we need to look for different choices (and share the results with the 
community..)

bye

andrea venturi


> On Tue, 2003-04-22 at 08:54, Andrea Venturi wrote:
> 
>>hi,
>>
>>we'd like to set up an asterisk server on a lightweight x86 pc with only 
>>USB port.
>>
>>we found this cheap isdn adapter: winbond w6694a
>>
>>   http://www.winbond.com/PDF/sheet/W6694a.pdf
>>
>>this seems an actually unsupported by linux usb adapter but it seems 
>>*bsd supports it 8-)
>>
>>   http://home.c2i.net/hselasky/isdn4bsd/privat/temporary/
>>
>>it's passive card so not a plain ACM compatable!
>>
>>the datasheet seems quite complete WRT register description so i think 
>>it should be not a though task to try to support this card _but_ i have 
>>a concern.
>>
>>actually there are at least three (or four) environment WRT isdn on linux
>>
>>1- the classical isdn4linux effort (the hisax stack) but it's quite poor 
>>WRT ISDN signaling if you use the /dev/ttyI* interface and "TE" mode only.
>>
>>2- the newer capi approach (i heard it should be the default in kernel 
>>2.5.x and onward but i don't know if it's true today)
>>
>>3- the zaptel/libpri way: do you think it could be easy to leverage the 
>>tormenta skeleton? is the ISDN PRI signaling really the same ISDN Q931 
>>(the D protocol on the BRI signaling channel in euro-land..): i see that 
>>there's just a different module for the E400P and the E100P.. support 
>>this new ISDN card could be just as easy as develop this low-level module ?
>>
>>4- the isdn raw device from http://isdn.jolly.de/ : i know this is not a 
>>well known and developed kernel support but this has the advantage that 
>>you can have both NT & TE support for the hfc based card.
>>
>>IMHO the best approach could be number three (kudos to mark and co.) 
>>because:
>>
>>- it's fully GPL
>>- it's reliable and running _now_ (when will have a capi kernel support 
>>uncluttered !?)
>>- it's feature-full (and not poor like the /dev/ttyI* emulation)
>>- it's both voice & data ready
>>- it's both NT & TE (in PRI level this is a quite light difference.. 
>>maybe a PITA in the BRI level..)
>>
>>on the other end, it's not a well known environment outside the asterisk 
>>community so it's quite a challenge WRT well established "telco" driver 
>>in linux world..
>>
>>what do you think about the topic ?!
>>
>>bye
>>
>>andrea venturi
>>
>>_______________________________________________
>>Asterisk-Dev mailing list
>>Asterisk-Dev at lists.digium.com
>>http://lists.digium.com/mailman/listinfo/asterisk-dev