[Asterisk-biz] T1 Technology and VoIP Gateway Primer

[Matt Roth]

Asterisk Users / Asterisk Biz List Members,

About a week ago I cross-posted a message titled "Large Asterisk Setup 
(~500 Concurrent Calls + Scalability)" to Asterisk-Users and 
Asterisk-Biz.  For reference, the threads generated by that message are 
archived at the following locations:

http://lists.digium.com/pipermail/asterisk-users/2005-April/102823.html
http://lists.digium.com/pipermail/asterisk-biz/2005-April/004590.html

First, I would like to thank you all for your excellent suggestions and 
contributions. My misunderstanding of the Digium quad-span card's 
scaling limitations was corrected (PCI bus traffic is not the problem, 
it is the number of interrupts generated by the Zaptel drivers) and I 
was directed to replace the Asterisk Slave Servers in this diagram 
(http://home.comcast.net/~mroth01/LargeAsteriskSetup.gif) with a VoIP 
gateway.

Following up on that suggestion, I began researching T-1 time division 
multiplexing in order to understand where the DSP load on an Asterisk 
server originates and the best options for purchasing a VoIP gateway to 
offload that processing onto. The results of that research can be found 
at the following links:

T-1 Multiplexing - PSTN Side (http://home.comcast.net/~mroth01/T1-PSTN.gif)
T-1 Multiplexing - CPE Side (http://home.comcast.net/~mroth01/T1-CPE.gif)
Basic T-1 Time Division Multiplexer 
(http://home.comcast.net/~mroth01/T1-TDM.gif)
Telephony Glossary 
(http://home.comcast.net/~mroth01/telephony-glossary.html)
Sources (http://home.comcast.net/~mroth01/sources.html)

My understanding of the T-1 TDM and the PSTN side is pretty solid, as it 
is mainly based off of Intel Corporation's T1/E1 Technology Primer (see 
Sources), but the CPE side is largely deduced from what I knew about the 
PSTN side.  There may be holes or mistakes, so I would appreciate any 
corrections or additions that you can offer. Specifically, I would like 
a detail of the TDM - VoIP conversion process, similar to the basic T-1 
TDM one I provided.

The differences between a T-1, DS-1, and ISDN are subtle and not 
universally agreed upon.  For a discussion of these issues see the 
following links:

What's the diff between a T1 and a DS1 
(http://pbxtech.info/showthread.php?t=1100)
PRI setup (http://pbxtech.info/showthread.php?t=1250)

In closing, I have a few questions:

- Is my understanding of using the same codecs and signaling protocols 
on both sides of the Asterisk server in order to circumvent transcoding 
and conversions on the server correct?

- Are there any other host-intensive processes that I should consider 
offloading to the gateway, such as echo cancellation?

- What does the PCM µ-law codec used in T-1 multiplexing map to in terms 
of Asterisk codecs (G.711 µ-law, perhaps)?

- What codec does the Monitor application use when digitally recording 
calls (if possible, I would like to avoid transcoding the streams when 
recording and let sox handle the conversions on a different box)?

Thank you for your time,

Matthew Roth
http://voip-info.org/tiki-index.php?page=Running%20Asterisk%20on%20Debian