[Dundi] representation

[John Todd]

At 11:22 AM -0500 on 10/26/04, Mark Spencer wrote:
>>Both +1700 and +1800 seem like parts of the E.164 number space 
>>which do not fit well within any system which seeks to answer 
>>questions like "where should I terminate a call to address B?". 
>>Maybe what is needed is an extension to provide answers for "where 
>>should I terminate a call to address B from address A?".
>
>what about 888, 866, 877 etc.  does that hold true for those numbers as well?
>
>If we have to use callerid, that could be very damaging to the cache 
>if not properly implemented.
>
>Mrk


I agree, but I would say that it should never be implemented in the 
e164 GPA-enforced peering mesh.  If we cannot assume that e.164 
number represents unique ENDPOINTS, then we've kind of lost control 
over any sort of hope for a global numbering space that is coherently 
routed.  Origin of call cannot possibly regulate where a call ends up 
in a public e.164-based inter-organization routing mesh.  Once a call 
gets inside of a particular ITAD (Internet Telephony Authonomous 
Domain - yes, I'm going to start using this term, since it's cleaner 
than "organization" or "company" or "authority") then that ITAD can 
route based on ANI or IP address of transmitter or whatever else they 
want - I have no objections to internal policy.  But external policy 
and routing announcements must be consistent across all those who 
want to participate in the same "global" route exchange.  It just 
won't work, otherwise.

The fact that North America has in the past (currently?) allocated 
non-geographic toll-free numbers in a fashion that is insane does not 
mean that we should be insane for all time to come by implementing 
insane exceptions in DUNDi or any other route protocol.  Richard or 
someone else more knowledgeable about ITU e.164 specifications can 
correct me, but I was under the impression that one of the parts of 
the specification were that destination numbers were to be unique and 
non-overlapping in any portion of the number space.  The 
determination of "unique" means that there is a registrant in the ITU 
tables, and also in the national tables at _the_ regulatory authority 
for that nation.  Seems pretty obvious, right?  Even if there are 
various announcements by "carriers of record", "service providers", 
and "end users", there is a clear legal responsibility chain across 
all three entities which can be followed up to the ITU in a single 
straight line.  This is what is needed.

Even if we have other nations where origin regulates destination 
(some parts of Eastern Europe, I'm sure have this problem) that 
SHOULD NOT CHANGE THE WAY ROUTES ARE ANNOUNCED in e164 in DUNDi. 
Look, if N routes have to be announced with N^N variations for 
different routes for different origins, we should pack our bags and 
go home.  Origin has NOTHING to do with destination in a telephony 
routing system where the assumption is free calling.

The mistakes and horrible hacks that have been done to numbering in 
North America, with our impossible legacy of inclusions of no less 
than 5 nations into a single "country" code, should not be a monkey 
wrench in a routing protocol.  Simply because NANPA toll-free numbers 
are broken (but may not be now?) please don't let this break DUNDi or 
any other routing protocol.  We're getting over our geographic 
attachments and ANI-based silly behaviors here; let's not allow these 
shriveling portions of our past mistakes to halt or fundamentally 
change progress with DUNDi and other protocols.

Simply because someone notifies NANPA or any other 
national/international numbering authority about use of an "area 
code" does not mean that area code is for their use, regardless of 
their good intentions.  It's _not_ an e.164 number, regardless of who 
they notify, and it shouldn't be in the "e164" context in DUNDi.  We 
need a non-geographic numbering space here in North America, since 
the +87810 space seems at this point to be a wasted effort despite my 
previous high expectations.  Discussions were had at VON which gave 
me hope on this front; if I'm successful in making any headway this 
list will be one of the first to know.


Please note: I am not suggesting that it is undesirable to have 
different costs based on origin or other criteria, and it may be 
useful for DUNDi to be able to represent different destinations based 
on origin (though I can't quite see how that would work...)  In fact, 
I am a big proponent of a routing system that understands monetary 
costs and can represent various different cost cases, perhaps based 
on origin.  But that's neither something DUNDi can handle right now 
nor is it what we're talking about, which is the free interexchange 
of routes in the GPA-enforced e164 context of DUNDi.  In any case, I 
don't think that even a cost-based routing system would have 
different route destinations based on origin; I think the only thing 
that would change would be the price or QoS or something else - the 
same phone would ring regardless of where the origin was.  But that's 
a chapter from a different book...


JT