On Aug 16, 2011, at 10:34 AM, Ted Lemon wrote:
> On Aug 16, 2011, at 9:43 AM, Fred Baker wrote:
>> In both cases, the addresses are concocted by the system using them. For RFC
>> 4862, that means "when the system receives a new prefix in an RA". My
>> understanding is that Windows privacy addresses are generated daily and held
>> for a week; I would expect other implementations to do something akin to
>> that. I wouldn't not expect DHCP servers to be doing updates in a world in
>> which the end host calculates its own address.
>
> The DHC working group is working on a draft that leverages DHCPv6 to allow
> hosts to register SLAAC addresses, CGA addresses and other self-generated
> addresses with the network infrastructure. Implementation would obviously
> be optional, but it's a good way to solve the problem—you can actually do it
> with DHCP already, but the mechanism hasn't been formally described or
> recommended.
>
> I would expect client-side implementations to include a switch allowing the
> user to disable this functionality, but it is a useful way to do things like
> populate the reverse tree and populate/update AAAA records for clients that
> want or need this capability.
Seems reasonable. It has the same authenticity issues as DDNS; the fact that a
system with an address announces a name doesn't make it so.
_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www.ietf.org/mailman/listinfo/dnsop
