Re: [dnsext] RFC 2142 and "organization's top level domain"

[Kevin Darcy <[e-mail]>]

On 9/10/2010 8:48 PM, Dave CROCKER wrote:
> Folks,
>
> I'm trying to resolve an erratum, on RFC 2142, MAILBOX NAMES FOR 
> COMMON SERVICES, ROLES AND FUNCTIONS".  The current focus is a narrow 
> matter of nomenclature.  Broader issues about a full revision effort 
> for the document are worthy, but outside of the current scope.
>
> RFC 2142 uses the phrase "organization's top level domain".  This 
> refers to the string that is delegated to the organization and 
> provides the root of any sub-tree to the DNS that they create.  Around 
> the world and around the DNS, this 'root' will have two or more 
> fields, such as bbiw.net or ucl.ac.uk. (And with recent TLD rule 
> changes, presumably this now is /one/ or more fields...)
>
> So it would refer to ietf.org, rather than tools.ietf.org or 
> www.ietf.org, and ucl.ac.uk, rather than www.ucl.ac.uk or 
> bartlett.ucl.ac.uk, for example.
>
> The construct of a domain name that an organization registers, and for 
> which it receives a delegation, is important.  Yet there appears to be 
> no common term for it.  RFC 2142 was written quite awhile ago.  It 
> uses a term that is formally reasonable, but which has become 
> potentially confusing.  The reference to "top level" for a domain 
> name, is now so universally taken to mean a TLD, that it is reasonable 
> to assume that the "organizational" qualifier is not sufficient to 
> move the reader away from the meaning they are likely to expect.
>
> So the question is what other term to use?
>
> I raised this question to the IETF applications discussion list and we 
> wandered over various choices.  Tony Finch came up with one that I'd 
> like to declare the winner, unless you folks come up with something 
> better:
>
>    organization's principal domain name
>
>
> Comments?
"Principal" is too subjective, in my opinion, since it can be defined as 
"most valuable" or "most important", which is in the eye of the beholder.

Given the context and apparent intent of the RFC, there should be some 
hierarchism in the term, so that [e-mail], for instance, would 
cover all of the subdomains beneath that registered/delegated domain, 
e.g. smtp.example.com, www.example.com, even if "example.com" itself may 
own no A records and thus may not, through, say, a "web"-oriented lens, 
be viewed as "most valuable" -- since it generates no hits, lures no 
eyeballs -- and thus, arguably, is not the "principal" domain name for 
the organization.

But, how to capture that hierarchism, without using the lately-confusing 
term "top level", and without impeding members of the intended audience, 
who don't happen to be DNS experts, with alternative yet bulky verbiage 
about the namespace hierarchy (e.g. "common ancestor", "highest-level 
enclosing", "least-labels enclosing")?

I don't know that it's possible to retain hierarchism without defining 
the relevant words/terms from the DNS realm, but if defining-verbiage 
cannot be avoided, the jargon term "apex" might be shortest and crispest 
one to use, as in "[organization's] apex domain". It would only need to 
be defined *once*, of course, and that would cover all 3 uses of the term.

                                                                         
                                                                         
                 - Kevin