North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

ICANN, TLDs, and directory services.

  • From: rdobbins
  • Date: Fri Nov 24 12:43:54 2000

I'm no big fan of Brock Meeks - he strikes me as being rather immature and
naive much of the time.  Still, his piece on the ICANN TLD meeting (see
http://www.msnbc.com/news/493721.asp?0nm=-11P ) is certainly enough to make
those of us who've been around for a while want to weep and gnash our teeth.

The first taste of this sort of nonsense was, in my opinion, the CLNP fiasco
of a few years ago.  Fortunately, wiser heads prevailed, but the blatant
power-play that certain parties made at the Tokyo (it -was- Tokyo, wasn't
it?) IETF meeting seems in hindsight to've been a foreshadowing of the
bureaucratic diktat under which we now seem to be operating.

ARIN's ill-informed (hadn't anyone there heard of SSL?) policy announcement
regarding virtual hosting IPs was yet another - hopefully, this will go the
way of the dodo after the public thrashing they've been subject to in this
and other forums as a result of said stupidity.  Still, the root problem
(pardon the pun) of the namespace and how it should be managed is a pressing
issue, and one which hasn't been resolved to anyone's satisfaction, not even
ICANN's.

IPv6 will, I think, begin to seep into mainstream usage as a matter of
necessity.  It's taking longer than it should, but security concerns alone
(i.e., the fact that, as far as I can tell, spoofing addresses under IPv6
seems to be a pretty difficult, if not impossible task) will probably help
give it some momentum.  I wonder how long it will be until the 'research
only' caveat will be eliminated by the registrars?

The namespace is the big enchilada, though.  There's been talk since at
least 1993 of either a) switching to a directory service as the naming
system for the Internet, or b) converting DNS itself into a true directory
service by simply adding more and more directory-like features.  X.500 was
what most people saw as being the logical choice; its cumbersome design and
the fact that it took so long to get the standard itself up to snuff made
that choice stillborn during the 1990s.

Now that we're starting to see, in very blatant terms, what
technical/policy/regulatory issues arise with the current naming structure,
I wonder if taking a serious look at directories as an ultimate DNS
replacement might be more than just a quixotic thought-exercise?  Structured
properly, probably along national lines, the TLD/top-OU issue would go away,
leaving people free to structure their namespaces as they choose.  Object
types and properties would certainly be a more elegant way of identifying
various content-types rather than merely relying upon URL prefixes plus file
extensions and their (assumed) proper mapping to MIME-types on the millions
of client machines around the world.

There are standards issues which must certainly be addressed before such a
change could be contemplated; even so, I'm beginning to wonder whether or
not it might be time to revisit the naming issue, with a specific focus on
directory services?

-----------------------------------------------------------
Roland Dobbins <[email protected]> // 818.535.5024 voice