[dns-operations] Alternate IPv6 address to hostname mapping, Re: IPv6 PTR records

Shane Kerr shane at isc.org
Wed Dec 22 13:02:57 UTC 2010


On Fri, 2010-12-17 at 16:45 +0000, Paul Vixie wrote:
> noting, bill simpson tried to get us to abandon PTR for IPv6, he wanted to
> just add a new "get hostname" ICMP message and let a host answer for itself.
> nobody else thought this was a good idea because it's a change to the security
> model (we knew DNSSEC was coming, and we thought ICMP would never be secured)
> and also because of the need to do address->name lookups while the host itself
> is not online, like once-a-day syslog postprocessing and similar tasks.  but
> i'm still not sure we made the right choice.  PTR has an impedence mismatch
> with IPv6, and IPv6 is the future of the internet.

I thought a simple ICMP message for this was a brilliant idea. I vaguely
recall that network administrators hate this because they think it pulls
control of the network out of their hands. And that DNS people hate this
because it pulls something out of the DNS.

To be honest I don't think there is anything wrong with defining two
separate ways to convert an address to a name, especially given that the
operational reality of IPv6 makes it a tricky problem.

I think all reverse DNS is stinky. It just stinks 96 bits more in IPv6,
so the smell is harder to ignore. ;)


More information about the dns-operations mailing list