[Asterisk-Users] Re: 4 port BRI options ?

[Louis-David Mitterrand]

On Fri, Jun 03, 2005 at 02:39:48PM +0100, Gavin Hamill wrote:
> On Friday 03 June 2005 14:28, Nardis Dome wrote:
> > --- "Brett, Gary" <gary.brett at cetelem.co.uk> wrote:
> > > Is the Eicon that much better ?
> >
> > sorry, i have only experience with Eicon... maybe
> > someone else is able to give a feedback...
> 
> Aside from paying for a recognised brand name, with Eicon you get on-board 
> DSPs and firmware, as well as software to make the card appear as a series 
> of /dev/tty devices (and/ or CAPI) under Linux.

With kernel 2.6 you only get ISDN tty devices through the capidrv
module, no more analog modem login support with the obsoleted diva2i4l
module.

> The Junghanns BRI boards have no such on-board logic, they are pure telephone 
> interfaces driven by the host - and as such they are /perfect/ for Asterisk.

You get benefits from the zaptel interface (zapscan, zapbarge, etc.),
closer to the metal, less latency.

> We use an Eicon Diva Server 4BRI for our fax server because of the TTY 
> interface for Hylafax - the DSPs themselves do all the fax negotiation / 
> compression and it works extremely well.

It works, but when ISDN channels are busy faxing, asterisk has no way to
know which outgoing channel is free: users get a congestion when trying
to call out (unless you program some channel testing logic/loop inside
your dialplan). This is a major bummer in production.

> However, we're also running * on that machine via CAPI, and the board
> gives us the flexibility to share 8 channels between * and Hylafax
> really easily :)

There is no sharing, they contend for the channels, even thought they
both use the capi layer. Go figure.

My experience with both cards is that the Diva is a waste of money. It's
better to go with the Junghanns card and do faxing from inside asterisk,
which works very well in the latest versions.

-- 
They can have my jeans, as soon as they pry them off my dead, cold ass.
-- (Spackler on /. about corporate dress code becoming formal again)