[ppml] Draft ARIN Recomendation

Thomas Narten narten at us.ibm.com
Thu Oct 21 20:49:49 EDT 2004


Some comments.

First, I must confess a bit of surprise at the reaction during the
meeting and here on this topic. This draft has been around for a long
time (a year in the making?) and I presented much of the same content
in Vancouver; there was no such reaction then.

Personally, I have to wonder if having ARIN take a formal position is
the way to go here. First, it will take some time to reach such a
consensus, if indeed, there would be consensus. The unique-local-addr
document is undergoing IESG review now; any comments need to come in
soon, not a month or two from now.  Second, this would be a first (I
can't recall ARIN ever having done so), so you'd be charting new
ground. Does the issue here warrant it?  Finally, what will get the
IETF's attention more than anything is clearly articulating what the
issues are, and how they can be fixed, if indeed they can. Saying "bad
idea" without saying why isn't particular constructive.

Also, please read the actual drafts at issue and be specific. In one
conversation I had after the meeting, Leo and I seemed to have
agreement that the two documents:

    draft-ietf-ipv6-unique-local-addr-06.txt
    draft-ietf-ipv6-ula-central-00.txt

are not equivalent. The first document doesn't result in guaranteed
uniqueness, so one issue that will come up (if ISPs start routing
these on a large scale) is what happens when two sites pick the same
prefix. Who actually "owns it"? Because there is no clear answer,
there may well be less incentive to carry such routes.

On the other hand, ula-central does call for centrallized allocation
of prefixes, so there is a much stronger binding between an end site
and a specific prefix.

So, do folk feel like both documents are equivalent in terms of "bad
idea"? And why?

Thomas



More information about the ARIN-PPML mailing list