[Asterisk-Dev] Business Case for a DS3 Channel/Driver in Asterisk VoIP

[Mike Machado]

> Because the phone companies still run the phone network.  For sending 
> 600+ calls over a circuit to a customer, they use DS-3s.  Ipso facto, if 
> you want 672 channels from the phone company in your Asterisk box, you 
> need a DS-3 interface.  Otherwise, you must either split the DS-3 up 
> into DS-1s or provision more DS-1s.  The monthly local loop charges (the 
> cost for the ILEC to give you copper from the MDF to the demarc) alone 
> for 28 PRI DS-1s is probably more than all the (home) phone bills you've 
> ever paid in your life combined, assuming you've been paying a phone 
> bill for 20 years or less.


How practical is this though? I had a 2.4Ghz P4 server with 1GB RAM and
a TE410P and I could not even run 4 PRIs at full capacity. I think I
made it to about 3 before D channels started bouncing. Not sure of the
overhead difference of CAS vs CCS.

I also had a 600 Mhz Mini ITX box with the TE410P and could not get past
2 PRIs.

Im sure something could have been done to optimize those configurations
a bit, but it still seems that scaling from 4 to 28 T1s is quite a jump.

I have nothing tangible, but simply multiplying that by 7 seems like you
would need one beefy server, beyond whats probably practically available
today.

Or am I way off here? If so, can you give examples of what kind of
hardware would accommodate this setup, CAS and CCS. PSTN to SIP,
assuming no codec translation.