To:
Kevin Darcy <kcd@daimlerchrysler.com>
Cc:
dnsop@cafax.se, comp-protocols-dns-bind@moderators.isc.org
From:
Shane Kerr <shane@ripe.net>
Date:
Fri, 8 Jun 2001 10:47:48 +0200
Content-Disposition:
inline
In-Reply-To:
<3B201B46.97FDF5A4@daimlerchrysler.com>; from kcd@daimlerchrysler.com at 2001-06-07 20:24:38 +0000
Sender:
owner-dnsop@cafax.se
User-Agent:
Mutt/1.2.5i
Subject:
Re: Multiple PTR records
On 2001-06-07 20:24:38 +0000, Kevin Darcy wrote: > > I wonder if this would be good BCP material (?). RFC 2181 (not a BCP > of course but Standards Track) almost seems to *encourage* multiple > PTRs by "clarifying" that it is supported in the protocol. Now that > the cat is out of the bag, perhaps there should be a BCP stating that, > while multiple PTRs are technically possible, they are generally > undesirable and when taken to extremes can in fact cause problems. > > I would not volunteer to write such a document, of course, given my > even-more-radical view that reverse DNS should probably go away or its > use be severely limited (and I don't think keeping reverse DNS around > solely as a sort of "ISP intelligence test" is really a strong > argument, even when couched in terms of spam-prevention). This is an intriguing idea. To be honest, at first glance it doesn't seem *too* radical to me. I mean, what's the real "use case" for referse DNS? Certainly any "match forward/reverse" for "security" has long been discredited. The ability to put a machine name in my utmp entry seems like a very small gain for such a large system. OTOH, my understanding of the IPv6 world is, "yes IPv6 numbers are totally ridiculous, so use DNS for everything". In such a world, reverse DNS seems to take on a huge importance. Not that I've heard any proposals how ISP's are going to manage running reverse DNS for the /48 they're going to have to give out to each dial-up customer. (I'm sure that somebody has a clever solution for this, I just haven't heard it.) -- Shane