[asterisk-users] Please some enlightment on ENUM !!

[SIP]

John Novack wrote:
> Raimund Sacherer wrote:
>   
>>> Adding random digits to a PSTN and expecting to get the same person at a
>>> different extension....  you don't think that's a hack? I do. One should
>>>     
>>>       
>> Sorry, please do not call a whole country using a hack when their solution is
>> legitimate.
>>
>> Austrian PSTN
>> https://supportforums.cisco.com/docs/DOC-1343
>>
>> Excerpt:
>> Dialplan
>> The Austrian dialplan is a variable length numbering plan, which consists of area codes and subscribers numbers. Area Codes are from 1 to 4 digits in length, subscriber numbers vary even more. Some Numbers (mostly used by companies) allow the use of Direct Inward Dial Extension. If such a number is used, it is up to the company to decide on the length of extensions, meaning that any length can be used.If too long extensions are used, the numbers might not be reachable from some sources. This is especially relevant with regard to number presentation inside of isdn, as stated under "Voice Interfaces and signalling".
>>  
>> A copy of the Austrian dialplan is maintained by the austrian regulatory council at http://www.rtr.at/en/tk/nationaleRufnummern.
>>
>>
>> The emphasis on "variable length numbering", if you consider that austria is a small country which is physically not able to harbor more than 16 million peoples in a sane way, the numbering plan is more then sufficient, austria is not like germany with > then 88 million inhabitants which needed a reconfiguration of their numbering plan some years ago.
>>
>> my 2cents
>>
>> Raimund
>>   
>>     
> For those who have no background in telecom history, this may seem 
> strange, but in fact in bygone years this was not JUST Austria that had 
> this scheme. In the Electro-mechanical days what some of us know as 
> Direct Inward Dialing ( not the DID term often misused in modern times ) 
> was handled this way in open numbering plans.
> It is unfortunate, but all too common, that in a great many fields,very 
> smart educated people are ignorant of the history of their field, and 
> are doomed to re-invent the wheel, or proceed down a blind alley.
>   

By the time telephone operators began to be replaced by mechanical
switches, open numbering plans became impossible to design for. Once
software switches came about and massive modifications to numbering
plans became as easy as coding new exceptions and pushing them out to
the millions of switches on the network, numbering plans had, largely,
been codified to make for logical and understandable patterns.

> In North America, with a closed numbering plan, all numbers are of a 
> fixed length, 10 digits. Technically the "one" is NOT part of the 
> number. In earlier days, no "one" was needed even for toll calling to 
> distant cities and area codes. Some, fewer each year, are able to dial 
> within their NPA with 7 digits. with the NANP turned over to the inmates 
> ( the state PSC's ) some locales require 11 digits for all calls, others 
> 10 for local, 11 for toll, and others 10 digits for all calls.
> The closed number plan is somewhat easier to parse, with its fixed 
> length, and no timeout or send/end digit is needed. the open plan can be 
> more efficient in use of numbers. Different locale dialing pattens do 
> make that more of a challenge, however.
> How all of this works in regard to ENUM is for others to decide, but if 
> it cannot handle an open numbering plan with variable length numbers it 
> needs fixing.
>
> John Novack 
>
>   
ENUM has no issues with variable length numbering plans in its design.
However, you have to stop calling it ENUM if it gets above 15 digits,
since it's no longer a valid E164 numbering scheme as per the design
there. ENUM is even applicable to local-only dialing plans, in which
you'd run your own server, point your phones to that server as their
primary ENUM server, and off you go.

It's a very flexible idea as far as mapping E164 numbers to services go.
However, once you get into the realm of registering with the IANA
approved servers and trying to place your own ENUM DNS server into the
mix, you can't start mixing local-only and public numbering schemes or
things break.

Austria is somewhat of a special case in which their numbering schemes
are such that they allow the ad-hoc creation of additional virtual DIDs
by simply tacking on digits to a valid DID. It's an open numbering plan,
but from what I gather, it's a variation of the traditional open
numbering plan in that each DID owner or designate gets to create his
own additions instead of the telco approving all variable-length DIDs.
This doesn't break any ENUM rules (unless the number exceeds 15 digits),
but it does create a scenario in which it may become difficult to apply
traditional ENUM tools to the scenario at hand with an attempt to get
the results you're after.

For instance, if user X owns the number +4311234567, and he decides he
wants to create a slew of virtual DIDs after that (+4311234567[01-99]),
it doesn't violate the ENUM standard because all those numbers are fewer
than 15 digits, and therefore all valid E164 numbers.  But the question
was how to register them and if user X should create his own DNS server
to handle the ENUM mappings. User X is free to do this, but then how is
that server reached to map those numbers? There are certain 'accepted'
ENUM registrars that have been IANA approved. User X could attempt to
get his own approved, but even then, most existing ENUM tools would
still not know of its existence, since it would be new.

Options are (as were provided) to register each of those numbers with an
approved registrar (bulk registrations are often available), or in some
way to convince a registrar to forward any and all requests to
+4311234567[XX] to User X's DNS server. THAT might be tricky, but it
creates a logical scenario in which there are 2 extra digits of a
variable nature to pass off to the DNS server of the registrar and then
forward to User X's DNS server. 

However, when you get into the idea of creating additional digits, it
may become tricky simply because of the way DNS handles requests.  No
longer do you have a simple *.*.7.6.5.4.3.2.1.1.3.4.e164.arpa mapping
zone, but you essentially need to create
*.*.*.7.6.5.4.3.2.1.1.3.4.e164.arpa, and
*.*.*.*.7.6.5.4.3.2.1.1.3.4.e164.arpa, etc, etc. just to handle all
cases. And I'm not even sure that would work other than in theory (I've
personally not tested such a scenario), so you're back to the fact that,
while useful for a country full of people, a dynamic numbering scheme
that does not correspond to standards cannot easily be handled by tools
which are designed to comply with standards.

And Austria's open numbering scheme is not standards. There are other
open numbering schemes in a handful of countries, but they behave
differently -- each with its own ideas on how they should be
implemented. If you want to 'fix' ENUM to handle that, you end up
creating a standard full of exceptions to handle the whims of random
telecoms throughout the world, and it no longer becomes a useful concept.

Again. This is my opinion. Feel free to have a different one. :)

N.