On 8 Aug 2005, at 21:16, Fred Baker wrote:

personally, I an see a *lot* of reasons to leave such decisions in the hand of the administration. The most compelling is: "try taking it out of the administration's hands. Just try. I dare ya."

DHCP/DHCPv6 seems like a reasonable choice of dynamic host configuration protocol.

Naturally I concur, however two technical challenges leap out at me concerning the nature of the policy table as specified by RFC3484.

First, the lack of field definition for labels has seen different OSes use different datatypes for the label, from string through stringified-integer to integer. Any cross-platform policy specification protocol would need to cater for this inconsistency. Perhaps this is something that a RFC3484-bis could address when making that spec compliant with the requirements of RFC3879 Section 4?

Also, RFC3484 permits the specification of zone index in the policy (c.f section 2.1). Section 6 of RFC4007 explicitly states that zone indices are strictly local to the node, making any centralisation of RFC3484 policy "challenging". [aside: OK, so administrators could harvest index data for links on which scoped-policy is required and then maintained in the DHCPv6 server as specific for that node, but there's no protocol for that harvesting and I've not surveyed sufficient implementations to determine whether indexes persist and are consistent between reboots]


So, yes, I agree that centralisation of address selection of policy is important (and necessary for folks using ULAs with greater-than- site scope multicast), and that DHCPv6 appears a reasonable choice, but there are fundamental issues with RFC3484 that need to be addressed at the outset.


Mark/

--
Dr Mark K. Thompson
Electronics and Computer Science
a School of the University of Southampton, UK


--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to