v4/v6 dns thoughts?

Jeroen Massar jeroen at unfix.org
Tue Aug 9 15:14:17 CDT 2011

On 2011-08-09 20:47 , Joe Pruett wrote:
> as i'm rolling v6 into my world, i'm not sure which way to go with
> reverse dns conventions.  for forward i'm doing things like:
> foo.example.com    a
> foo.example.com    aaaa    1000::
> foo.v4.example.com    a
> foo.v6.example.com    aaaa    1000::

You do mean:

foo.example.com       A
foo.example.com       AAAA  2001:db8::
foo.v4.example.com    A
foo.v6.example.com    AAAA  2001:db8::

I hope, seeing that is for the APNIC region and 1000::/8 is
outside 2000::/3 and thus not defined yet, that you use the
documentation prefixes when showing examples instead of abusing that
address space, as that is exactly the reason why will most
likely never be allocated to anyone but researchers who are seeing all
kind of fun backscatter...

> so i can use a foo.v4/v6 hostname if i need to specify transit behavior.

People commonly use the 'ipv4' and 'ipv6' variant for this. Most
network-specific tools though fortunately have -4/-6, but as indeed
quite a few don't it is always handy to have the above.

> being able to just use foo.example.com for authentication purposes
> (sendmail, nfs, etc) is nice.  but also knowing when incoming is v4 or
> v6 by just looking at the dns lookup (for tools that do reverse lookup
> for you) is also nice.

Tools that do reverse lookups should always also report the IP address
as without the IP a reverse is futile unless said tool does at least a
ip->reverse->forward check and then of course the hope is that that
hostname does not disappear between that lookup happening and it going
away again...

> what are you doing?  which way makes more sense to you?

Map it to the hostname. This as it should not matter if it is IPv4 or IPv6.

For routers of course one might want to use a v4/v6 specific one as per
the above reason of 'easier for the eyes in traceroute', but on the
other side one could just as well use an IPv4+IPv6 per interface and
thus name them based on the interface


More information about the NANOG mailing list