Anybody behind a NAT64?

Daniel Roesen dr at cluenet.de
Fri Dec 7 13:39:15 CET 2012


On Fri, Dec 07, 2012 at 03:15:47AM -0800, Cameron Byrne wrote:
> i think the simple way to handle this is that a DNS server with a
> "standard view" is available to clients over ipv4 and a DNS64 view is
> available via an ipv6 address.
> 
> The network operator can control which view the customer see via the
> automatic network configuration.
> 
> If the customer is dual stack or ipv4-only, push the ipv4 address of
> the dns server.  If the customer is ipv6-only, push the ipv6 address
> with the DNS64 view/

This assumes the provisioning system knows in advance what the client is
capable of. Which is certainly not the case in residential CPE
scenarios, and probably also not in mobile scenario. Think of context
establishment timing. I'm not into the mobile stuff at all, so bear with
me if I go down the wrong road, but with DHCP you have the issue that
e.g. a client might first do DHCPv6 and later DHCPv4. So at time of the
DHCPv6 request, the server cannot know wether a v4 request will follow,
so no informed decision to make about wether to signal IPv6-transport
DNS recursors (doing DNS64).

IMHO, DNS64 behaviour should be opt-in from the endpoint.

Best regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr at cluenet.de -- dr at IRCnet -- PGP: 0xA85C8AA0



More information about the ipv6-ops mailing list