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

Paul Vixie vixie at isc.org
Wed Dec 22 16:03:12 UTC 2010


> Date: Wed, 22 Dec 2010 22:07:35 +0900
> From: Randy Bush <randy at psg.com>
> 
> > PTR has an impedence mismatch with IPv6
> 
> perhaps this deserves a bit of amplification

ok.  in ipv4 there's been a long standing problem in that the A/PTR model
for host name->addr and addr->name did not take broadband into account, it
came from a time when hosts were rare and the name/address bindings were
mostly static.  some work went into dns dynamic update + dhcp integration
but has had very little impact on most hosts which are broadband accessors.
since some internet transaction forms require addr->name bindings and since
there was no other way to create them, access providers depend on things
like $GENERATE to create a lot of pro forma PTR's just to have matching
A/PTR.  this was an unpleasant workaround but the design tolerated it.

in IPv6 the same pressures exist but that workaround does not work anymore.

if we were looking for a better host name/address binding model we'd want
to solve not just the IPv6 model problem but the old IPv4 model problems
that have dogged us for a long while.  however, i would like to see it
remain in DNS and not move to ICMP, for a lot of reasons that go beyond
shane's quip ("i'm a DNS person and i don't want to see anything taken out")
even though i do resonate to shane's quip.



More information about the dns-operations mailing list