[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

v4 given v6


At 8:59 AM -0800 1/23/98, Buclin, Bertrand wrote:
>Hi Bob,
>> Can advocates of maintaining this convention please state their reasons for
>> this to the list?
>As one of those using a separate domain for Ipv4 and IPv6, here is my
>(only) reason to do so:
>The IPv6 lab I am running is an experimental environment, and it is our
>corporate policy not to mix experimental stuff with production. I guess
>anybody around running a service business has similar policies.
>DNS use for IPv6 is far from mature yet, and if we want to experiment
>with the new proposals, we should not constrain ourselves with
>production-level requirements. We want to be able to upgrade the DNS
>servers quickly if new DNS RR types are introduced, and there is no way
>I can push Operations to put  on production nodes an early and un-tested
>BIND release...
>Also, several of the nodes in my lab only have IPv6 stacks and no Ipv4.
>So, trying to find what their Ipv4 address does not really make sense.
>As we will see more native transports for Ipv6 (over PPP, ATM, Frame
>Relay,...), more and more Ipv6-only systems and especially routers will
>be on the 6Bone.

Thanks for the reply.  Your points are well taken.

My main purpose is to enable clients to decide which stack to use by
querying the dns for a specific domain name, which seems to require (in
general) the A and AAAA record being under the same domain name "path".

Maybe an easier way to do this for now, rather than trying to have everyone
move their AAAA into the production directory, is to have the IPv4 A record
appear along with the IPv6 AAAA record in the experimental IPv6 subdomain.

Regarding the state of most current BIND releases in common use, it was my
understanding that AAAA support way widely implemented (even if the site
doesn't know about it :-)   Anyone know if this is remotely true, or am I

I appreciate this wasn't your point on newer RR types in advanced BIND

>The initial question was about how to find automatic tunnel end-points.
>Even if one find which is the Ipv4 address of my boxes, that does not
>mean that I am ready to offer an automatic tunnel on those. Actually, on
>almost all my systems, automatic tunnelling is disabled.
>What might be of better use would be to have a new application code for
>the registry letting people know the IPv4 address of systems accepting
>auto-tunnels (After all, tunnels are also a sort of application).
>Something like:
>application: Auto-Tunnel ::a.b.c.d
>application: Auto-Tunnel domain.name
>provided there is a AAAA (or whatever will replace them) for domain.name
>translating to ::a.b.c.d

I can't evaluate the value of doing this, so would prefer others comment,
but if there is a use in doing this we can ask David Kessens to add it.