Scott Howard wrote:

Clearly the end result is the same - my phone rings - but without ENUM or something similar there is no optimization of the path taken, and the call is almost certainly going to leave the TCP network and go through at least 2 SIP-SS7 translations - and with a much higher cost both to you and to me!

Let's say one in a hundred calls I make could by via Public ENUM. I pay about AU$3/hr to

That number could be so much higher - at least if the big players joined in. Based on http://gizmo5.com/pc/backdoor/ over 10% of all calls would be a definite possibility _today_, and that number is only going to get bigger over time.

call your US number. Is the cost of having a server setup at home to save AU$3 once in a hundred calls (for us that's a few months) really worth it?

As I said earlier, I don't see ENUM(*) as a home-user thing - it's a VOIP provider thing. How many calls go out of Internode (or MNF or Vonage or any other provider) today that traverse the PSTN network only to terminate on another VOIP providers system? At what cost?

Sure, if someone wants to come out with an ATA that does ENUM(*) then bonus to the people that use it - they get free calls to some people (probably along with several NAT-traversal issues), but the real win here is for the providers.

The real question here is whether any of the VOIP providers want to try and turn SIP-based VOIP into a real network, or if they just want to leave it as the technology for the end-points, with PSTN looking after the routing. Historically simply doing the latter made sense, but as critical mass changes towards VOIP then avoiding PSTN for inter-provider calls makes far more sense. Private peering is a start, but even that is still the equivalent of filtering much of the worlds email through UUCP because it's easier than using SMTP everywhere...

(*) And by ENUM I mean "ENUM or something that works towards the same goal", not necessarily RFC3761-based ENUM.

Scott.