[Asterisk-Dev] Is there a need for a DS3 channel/driver?

[Steven Critchfield]

On Thu, 2004-11-18 at 15:44 -0500, Mike M wrote:
> On Thu, Nov 18, 2004 at 01:44:13PM -0500, alex at pilosoft.com wrote:
> 
> > can't do voice 
> > or anything reasonable on a unchannelized DS3 card
> 
> By channelized do you mean you have n tx/rx FIFOs, one for 
> each channel, brought to you by embedded CPU and ASICs and
> Xilinx, etc?
> 
> By unchannelized do you mean a raw unstructured stream of bits?
> 
> Sometimes in the so-called unchannelized streams you are informed
> of where the frame boundry is in the bitstream.  With that information
> one can create soft-channelization.  Perhaps this is unreasonable for
> a DS3 however.

Well under the current Zapata methodology, they gave each DS0 it's own
kernel device. Of course once it reached beyond a certain number they
moved on to a different way of addressing each one. The point being that
it is possible to use a zapata device as both a data device handled by
the kernel and as voice handled by asterisk. 

While it is possible for someone to want a DS3 of just voice into an
asterisk machine, it is also likely that they would be doing something
like VoIP termination and wanting to have some number of DS1s as data
and some as voice and therefore a soft channelized DS3 might not be very
usable if the split is done outside of the kernel unless someone wanted
to use the ZapRas type service to connect some number of channels to a
PPP daemon to handle the routing. Of course that would limit your
options for terminating the data on the other end.

So the question still comes down to what is available via the DS3 cards
drivers and whether they match up with what needs to be done with
asterisk. Combine it with the licensing and needs for either GPL or
special licensing fee from Digium.

-- 
Steven Critchfield <critch at basesys.com>