On Aug 16, 2011, at 7:27 PM, Brian E Carpenter wrote: Ted, all I meant by "political" is that it has been quite controversial in IPv6-land whether DHCPv6 becomes a de facto mandatory option, or whether sites can opt to use nothing but SLAAC for some or all hosts. If the conclusion for 6renum is that DHCPv6 is required, so be it, but the discussion might get noisy.
Hm. Okay, this helps me to see where you were coming from with your other point about garbage in, garbage out. If the problem you are trying to solve is that AAAA records of servers follow the server when you are renumbering, then you need secure DNS updates from the server that's being renumbered. But that seems to be an unlikely use case to me: typically I would expect servers to have manually-assigned addresses, or else CGAs, but not SLAAC addresses. Or perhaps addresses that are algorithmically assigned based on the prefix--prefix::1, as opposed to prefix::SLAAC. In any case, you do not need DHCPv6 to solve this problem, and so if that is the use case you are talking about, DHCPv6 is definitely not required. The case where you'd want to use DHCP is the case where non-server hosts are being renumbered. In this case, the name the host claims to have is a convenience, and arguably of limited value anyway. And so in that case I would argue that DHCPv6 is not _required_ either, although it may be convenient, depending on site policy. I wouldn't expect it to be used at a public hotspot, for instance, but could easily imagine it being used on a non-guest corporate network.
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
