Re: [v4v6interim] Single namespace

[Fred Baker <[e-mail]>]

fine. I'll argue that the IPv4 host in my example doesn't know  
anything about public or private addresses; it knows what server it  
has been instructed to use.

In the DNS64/IVI case, it knows little more. It knows that it has no  
IPv4 DNS server and no IPv4 router or address, and that it has all of  
those in IPv6. So it asks for a AAAA record for any name it needs  
translated into an address, and it believes the translation. The  
translation is scoped - the DNS64 server advertises AAAA records  
derived from A records when it has no other AAAA record to present,  
and it does so in the domain in which it resides. But the host knows  
nothing of the scoping.

On Oct 1, 2008, at 1:57 PM, Andrew Sullivan wrote:

> On Wed, Oct 01, 2008 at 01:42:11PM -0400, Fred Baker wrote:
>
>> Let's imagine that the name oatwillie.example.com has two A records,
>> pointing respectively to 1.2.3.4 and 192.168.0.1. The question is not
>> whether the name is different in different namespaces; the question  
>> is
>> whether the translation is different in different domains.
>>
>> The way we usually implement this kind of thing is to have multiple  
>> name
>> servers, one of which is accessible from outside and the other is
>> accessible from inside. That gives us one name space, but different
>> responses to that translation request depending on the source of the
>> request. DNS64 does largely the same thing, but commits the gross and
>> indecent act of saying out loud that it does so.
>
> I think that in the scenario with one RFC1918 address and one public
> address, one of the answers comes with a local context and the other
> comes with a global context.  The local context is determinable by the
> querying host on the basis of the host's own IP address, because you
> know that if you have an RFC1918 address, you're in a local-only
> context.  In the case of DNS64, however, the whole point is that the
> host doesn't need to know that it's living in a special place.  So
> there's nothing about the state of the host that tells you what the
> context is.  Instead, you need to know what the state of its relation
> to other hosts on the Internet is in order to know what the "right"
> context is.  Something about that makes me uneasy, but I can't yet put
> my finger on what.
>
> A
>
> -- 
> Andrew Sullivan
> [e-mail]
> [phone] x104
> http://www.commandprompt.com/
> _______________________________________________
> v4v6interim mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v4v6interim

_______________________________________________
v4v6interim mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/v4v6interim