[At-Large] R: China is going to divide the DNS
karl at cavebear.com
Sun Jul 1 22:38:18 UTC 2012
The real issue of that Internet Draft is not based in technology, rather
it is based in politics.
As many have pointed out, there is zero chance that the IETF process
will allow this draft to grow to any sort of internet standard status.
But the IETF is not the last word - it never really was. Just look a
the of network services as shown by IANA. It shows something on the
order of 15,000 network services many of which were done outside of the
context of the IETF -
There is no technical doubt that parallel, competing DNS roots could be
established. There are many who argue that that would cause a split in
the internet name space. It could. But that is a possible outcome, not
a necessary outcome.
Personally I look at the issue not as one of singularity or multiplicity
of DNS roots but rather as one of consistency.
Everyone, I hope, has seen the Monty Python tobacconist sketch - more
often called the Hungarian Phrase Book sketch in which a person has a
Hungarian-to-English phrase book that horribly mistranslates things.
Think of DNS roots as competing phrase books. A poor DNS root such as
shown in the sketch might be funny, but it would make its users angry
and would have minimal commercial prospects and, if it survived at all,
it would tend to evolve into a special niche.
Think of DNS roots as facing the same kind of pressure - if they are
inconsistent, i.e. if they surprise their users, then those users (or
their ISPs) will vote with their feet and choose a less surprising DNS root.
Now, there is the argument of misrepresentation - it is a valid
argument. But there are existing mountains of laws and regulations in
every country that can be brought to bear on people (natural or
corporate) that engage in fraudulent representations. It may be harder
than one likes to turn an accusation into a punishment, but due process
is neither always efficient nor always quick.
And there is a flaw in the internet architecture - which is the lack of
universal mutual identification and authentication. We tend to use the
internet as if we we though that every time we utter a domain name we
get perfect answers. Anybody who utters "google.com" in a web browser
while traveling learns that DNS names lack geographic uniformity. And
we all know that DNS names lack temporal uniformity because we have all
encountered DNS names that have been re-purposed.
Consequently that flaw in the internet architecture contributes to this
belief that domain names are somehow perfect master keys.
We would be silly to pick up a telephone, tap out a number we believe to
be that of our doctor and as soon as someone - anyone - answers we blurt
out our deepest secrets. We know better - it could be a wrong number or
someone else may have picked up. But on the internet we do not know
better, we blurt out like that.
So the problem with arguments about misleading data from competing roots
are based more on a lack of a universally deployed internet layer to do
consistent identification and authentication than they are based on DNS
My own sense is that if we allow competing roots we would not have
needed ICANN's TLD processes; new TLDs could have grown in much the same
way that new products aspire to shelf space in stores. Those TLD
products that got user acceptance would survive and those that didn't
would fail - that is true "bottom up" consensus rather than the rather
forced system we see in ICANN. For more on that idea see
Back to the internet draft:
In my business (testing of internet protocols for robustness) I see a
lot of corporate energy to create internet drafts in order to gain
ability to claim "we ware working within the IETF" while moving forward
on an idea no matter whether the IETF goes along or not.
I suspect that the authors of this draft are serious technologists who
are earnest about their ideas and that the ideas themselves are worthy
of examination and consideration.
But the larger political message is that the mantra of a singular,
rigidly catholic DNS is starting to evolve into a message that elides
the rigid hierarchy from exactly one provider to a message that
envisions something more like separate and equal hierarchies that are
sufficiently consistent with one another that users will not be
discomforted, at least not any more than they are today by client geo-IP
based name resolution.
More information about the At-Large