[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Alternative Roots Issue



In message <905DD86907DAD3119DE70000778D770F04E41246@mailsrv1.itu.ch>, "Androuc
hko, Vladimir" writes:
>Hello,
>Dear Protocol Council Members,

I'm not sure I'm responding to the proper version of this note -- I 
received three very similar notes, and two recall notes.  (It would be 
nice, I might add, if the "recall" notes cited the Message-Id of the 
message being recalled.)

Anyway -- we really need to rework the wording to make it clearer
and less ambiguous -- or at least less subject to willful misreading.
At the Montevideo meeting, John Klensin repeated the basic message
of RFC 2826 by pointing out that the formal, mathematical definitions
of things like "trees" and "roots" do not permit even a meaningful
discussion of multiple roots in the DNS.

Here is some possible alternative language, derived from Leslie's
earlier suggestion.

	The Internet currently operates using a tree-structured
	name space known as the DNS.  Of necessity, such a name
	space must have a single, authoritative root.  Moving to
	a model that would not require such a single, authoritative
	root would require replacing the present, working DNS with
	some other system.  Such a replacement would require the
	development of a new naming paradigm, as well as the
	protocols and software to implement it.  Developing and
	deploying such replacement protocols would take years, and
	would have enormous potential for disruption of the Internet.
	The PSO does not see any technical benefit in such an
	effort.

It says essentially the same thing as our earlier wording -- that
one can conceive of different ways to do name resolution -- but
points out the costs.  While there has been little explicit
discussion of the earlier, ambiguous, text within the IETF, some
who have seen it and the proposals based on it have reacted very
strongly.  For the Protocol Council to fail to take action in this
area would probably invite a unilateral response from the IETF.


		--Steve Bellovin, http://www.research.att.com/~smb
				  http://www.wilyhacker.com