July 21, 2004

Leaping Without Looking - And Taking the Internet Along for the Ride

Please note the "Update" at the end of this item.

ICANN announced with a great deal of precision that "[o]n 20 July 2004 at 18:33 UTC the IPv6 AAAA records for the Japan (.jp) and Korea (.kr) country code Top Level Domain (ccTLD) nameservers became visible in the root zone file with serial number 2004072000."

It might have been hoped that ICANN would have studied the issues pertaining to IPv6 in zone files before it allowed this to happen.

But it was not until 19:40 UTC, more than an hour after ICANN's announcement, when the IETF posted a new internet draft dealing with the issue:

        Title           : DNS Response Size Issues
        Author(s)       : P. Vixie, A. Kato
        Filename        : draft-ietf-dnsop-respsize-01.txt
        Pages           : 8
        Date            : 2004-7-20

In other words, ICANN rushed headlong into deploying IPv6 in the root zone even before the IETF published material on the subject, much less waiting for that material to receive peer review.

ICANN leapt before it looked.  And ICANN took all of us on the internet along for the ride, without our consent.  In my vocabulary that's called "reckless disregard of the consequences."

Contrast ICANN's behavior in this instance with the agonizingly slow review, a review that is now in its fifth year, of the question of introducing new TLDs.

It's my impression that ICANN was looking for a quick sucess story to mask its absolute failure in all other technical aspects of its job other than internationalized domain names.  I've read the new Internet Draft (ID) and run its DNS modelling program.  That ID answers many questions, but it leaves other questions unanswered.  For example, I believe that there is danger that the efficiency of DNS could be compromised because name resolvers might not be able to get as much "glue" address information as have until now and that resolvers may, as a consequence, end up engaging in previously unecessary name query interactions with root and TLD servers in order to obtain that information.  I could be totally wrong or the magnitude of the effect may not exceed some (never articulated) threashold.  But then again I might be right.  ICANN should not have moved forward until such questions can be fully articulated and fully answered.

Update: There was apparently a version zero of this ID published on the order of a year ago (June 2003).  Unfortunately the IETF has a policy of purging Internet-Drafts within a few months after they are published and as a consequence the prior version is unavailable from the IETF's I-D archives and has been officially unavailable for the greater part of the last year.  It is, however, visible at http://www.ietf.org/proceedings/03jul/I-D/draft-ietf-dnsop-respsize-00.txt In any event if ICANN can demonstrate that it knew of and, in fact, used that prior draft, then I can rightly be accused of overreacting.  On the other hand, even if ICANN did use that prior document there still remain those unanswered questions to which I alluded.  In fact, there is a CENTR presentation that alludes to these open questions, in particular see the third bullet on slide number 6.

Posted by karl at July 21, 2004 2:20 PM