[Inip-discuss] Domain Names

[Edward Lewis <[e-mail]>]

This message was originally sent to the iab mail list.  I'm repeating it
on this inip-discuss list to move the discussion to this themed list for
public discussion.

The original message:

As an outcome of work happening in the DNSOP WG, the work being the
addition of .ONION to the Special Use Domain Names registry and the group
looking to revamp the process for that registry, I researched the recorded
history of Domain Names and am working on a draft that can be found here:

https://tools.ietf.org/html/draft-lewis-domain-names-01

The draft isn't a completed work, its a start which I think needs
community review.  I'm looking to find the right place for it.  Despite
the seemingly natural fit to the DNSOP WG, I don't believe that is the
right fit.

I'm asking the IAB for it's opinion on whether this is a network
architectural issue or not.  I suppose the placement of the discussion
within the IETF is a matter for the IESG, so I'm not asking about that in
this message.  I only mention my qualms about DNSOP WG as a home as a
motivation to wonder if the problem is an architectural one or not.

What I've found, looking at the RFCs cited in the draft (as opposed to
IETF folklore, memories, mail lists), is that the idea of Domain Names
preceded the DNS.  There has never been a clear definition of Domain
Names, they have assumed the role of the basis for identifiers and are
used in many applications.  The DNS was built as an evolutionary
operational step, not as studied architecture for managing identifiers.

Is there a need to have an architectural understanding of how identifiers
are used across the internet's protocols?  The example of Tor's ONION
identifier (top-level name) pushes the point of how an identifier system
can follow the fuzzy definition of what a Domain Name is while yet
outright rejecting the management model of the DNS.  I'd once asked
(privately to a few, about 2 years ago) whether the DNS defines Domain
Names or do Domain Names define the DNS, in advance of this situation.  At
the time I don't think anyone understood my question.

I think, and I mean that as "think", there is a need to find a happy
medium where identifiers can be managed in different ways (DNS hierarchy
and zone admin vs. Tor's cryptographic basis) yet use similar syntax for
the purposes of sharing existing protocols (like HTTP) while using
different resolution processes.  The "victims" of not having a clear
direction are implementers of client applications that will have to juggle
different kinds of identifiers as well as those that implement lower (in
the software stack) layers that manage how an identifier is resolved.

Comments?