Immediate conclusions on iDNR

Jon Davis (jdavis@inetinit.org)
Fri, 4 Sep 1998 12:51:05 -0700


Message-ID: <001c01bdd83d$5ace0de0$637a0118@c603887-a.ptbrg1.sfba.home.com>
From: "Jon Davis" <jdavis@inetinit.org>
To: "URI distribution list" <uri@Bunyip.Com>
Cc: <ietf@ietf.org>
Date: Fri, 4 Sep 1998 12:51:05 -0700
Subject: Immediate conclusions on iDNR

Some conclusions have been made regarding iDNR, some of the more obvious of
which were already evidident in spite of my inability to communicate them
originally.  (iDNR is intended to be a radical alternative hostname
resolution scheme to DNS.)

Observation: DNS will never be forwards-compatible to iDNR.  The install
base of DNS is arguably equivalent to the install base of IP, so iDNR would
not be--and was never intended to be--appropriate for the existing install
base.

Observation: iDNR as it is implemented in a URI (or a URI-like format) is
not URI-compliant.  Because the install base of URI, like that of DNS, is so
large that it would not be--and was never intended to be--appropriate for
implementation on the existing install base.

Conclusion: iDNR is a niche protocol which is clearly incompatible with
existing standards.

 - However -

Observation: iDNR is intended to be backwards-compatible to DNS.

Observation: The as-of-yet-undeveloped specification in which iDNR names
would be implemented will be backwards-compatible to the URI spec.

Conclusion: The opportunity to start from scratch and reinvent (or evolve)
these heavily used protocols with alternative protocols, though with
different names and different standards, gives us the opportunity to
implement features that we have always wanted but which the older
specifications did not allow.  While the existing install base may not be
compatible, future implementations can target compatible systems.  (I for
one am willing to wait the five or ten years for large-scale
implementation.)



The while many specifications have been severely broken in iDNR, the purpose
of my bringing it to the attention of the IETF is to develop a specification
which has fewer limits and broader scope than those that the existing
install base uses.  It is my personal observation that DNS and even URI have
become obsoleted specifications by demand, but without an alternative &
evolved specification it would be impossible for an implementations entity
(i.e. software vendor) to make an upgrade when there is nothing to upgrade
to.

While the iDNR specification isn't really a specification at all yet but
merely a concept, there is much research to be done, including that of
implementation with LDAP, which has hardly been looked into.  Would it be in
the interest of the IETF to continue research of iDNR--outside of the URI
since it would not work with URI--in a dedicated mailing list?  Or perhaps
in an existing list?  Or, is the general consensus that of disinterest in
iDNR, and a waste of time and bandwidth brought on by a naive young dreamer?


Regards,
Jon Davis