[Asterisk-Users] VoIP extn number planning

[Rich Adamson]

Looking for some advanced thoughts relative to exten number assignments.

We're in the planning stage for rolling out asterisk at multiple small
US telco/isp operations. Their typical voip customer has had their 
pstn line for a looooong time and wants to keep the pstn line and number, 
but add voip to their existing home/soho arrangement.

One approach (from a planning perspective) is to deploy spa3k's at the
customer's location and configure it for pstn ring thru to line1,
implementing something like dial-8 for outbound voip calls, and
defaulting all other outbound calls to the pstn line. The arrangement
addresses 911 calls, etc, nicely.

However for inbound voip calls, there does not seem to be any industry
planning for addressing number assignments associated with voip 
facilities. (With 20+ years of telephony engineering experience, I do
understand the value of ss7, number portability, dundi, etc.)

On the surface there seems to be two basic approaches for these customers:
 1. If the customer's current pstn number is 402-234-5678, assign the same
 number to their voip facility, or,
 2. For each Central Office (in this case), assign 100 or 1000 local
 exchange number (from their existing pstn numbering plan) to voip
 customers, or,
 3. Assign some random number (something like FWD's approach) and translate
 that number within the voip switching facility (eg, asterisk in this case).

The voip number that is assigned doesn't make a lot of difference today
as the current spotty/independent voip-system implementations really 
don't address voip call completion on a larger/nationwide voip scale
(other than through existing pstn facilties).

The approach in #1 would suggest that each "remote" voip system must do
some sort of lookup on a per-number basis for call completion routing.
Keep in mind the exact same number is accessible via the existing pstn
network or via a voip network.

The approach in #2 would suggest remote translations might be easier
but probably does not support number portability in the long run.

The approach in #3 is certainly workable, but probably involves a number
change in the long term, disruption to existing customers, and in some
cases a truck-roll to facilitate the change. Not all that cool and
probably the highest cost in the long run.

Has anyone on this list given any serious thought on how to approach
this topic for the longer run?

Rich