[dns-operations] Vista implements a bizarre DNS server selection algorithm from RFC4384?

Joe Greco jgreco at ns.sol.net
Thu Mar 5 13:25:41 UTC 2009

> * Roland Dobbins:
> > Is anyone else seeing the described behavior?
> >
> > <http://drplokta.livejournal.com/109267.html>
> Yes, Vista isn't the first to implement this.
> There's an ongoing discussion on the IETF general mailing list and the
> namedroppers mailing list.  I don't think it makes sense to open a
> second thread over here.
> > I'm at a loss to understand the thinking behind this algorithm.
> There was a time when you were expected to get IPv6 addresses through
> a hierarchical addressing scheme.  RFC 2374 documents that approach.
> RFC 3484's Rule 9 makes sense in that context.
> Of course, it was very bad engineering to apply the same rule to IPv4,
> where addressing was anything but hierarchical when RFC 3484 was
> written.
> The IETF is structurally incapable of updating the specification,
> especially for IPv4.  It is considered out of scope for the DNS WGs,
> and the IPv6 WGs don't care much about IPv4.

I agree that by the time 3484 was written, we knew this wouldn't apply
to v4.  However, it is interesting to note that the design for v4 assumed
originally that there would be a smallish number of "tier 1" or "class A"
providers with huge allocations of space, and NAT wasn't a consideration,
and it might actually have worked within that context.

Many of us don't expect v6 to successfully implement anything resembling
2374 that will work in a manner useful to what we're discussing, because 
there's practical experience in the v4 deployment that shows it's naive.
I'm not saying that it's not theoretically possible, obviously it is, but
practice seems to be a different thing - just as we learned in v4.

There are DNS purists (hi Paul) who would want servers to return a fixed
set of RR's for any request, and I appreciate that.  There are good,
sound reasons to want to do that, as there have been in the many years
this subject has been hashed about.  On the other hand, there are a lot
of other creative things that get done as well, including lbnamed-style
answers (the opposite end of this problem!) ...

I just wish we'd start designing protocols and writing RFC's with the 
understanding that not everyone is going to adopt the sort of beautiful
and elegant strategy that an RFC author assumes.

That's my RFC vent-for-the-day.

Oh, and, I'm trying to remember, I know it was discussed at some point,
there are some practical strategies to discover the public IPv4 address
you're using on the Internet, was any of that ever collected into an 

... JG
Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net
"We call it the 'one bite at the apple' rule. Give me one chance [and] then I
won't contact you again." - Direct Marketing Ass'n position on e-mail spam(CNN)
With 24 million small businesses in the US alone, that's way too many apples.

More information about the dns-operations mailing list