updated address selection draft

[Tim Chown <tjc at ecs.soton.ac.uk>]

Hi,

I have compiled feedback into a revised draft available at:

http://tools.ietf.org/html/draft-chown-addr-select-considerations-03
 
These address comments on list, from Thomas and Dave, and include
reference to new drafts.   I believe the 6man chairs indicated
this could be a WG item, though I don't believe a list 'vote' was
called.   But this could be agreed in Stockholm.

General change notes:

- added reference to draft-denis-v6ops-nat-addrsel-00

- added reference to draft-arifumi-6man-addr-select-conflict-00

- noted that differing administrative domains are in scope (Thomas)

- noted that multiple interfaces are in scope (Thomas)

- noted node-wide problem is destination address selection (Dave)

- noted the two common multiple interface cases (wired/air, normal/VPN)

- noted any 3484 update has two elements: policy table and algorithms

- noted that many OSes already have implemented modified 3484 policy
  so we could use an improved 3484 'default' asap

- noted updates to policy not rapid unless node is in a site doing
  rapid traffic engineering changes or where nodes are heavily mobile
  between parts of the network with different policy (and thus actually
  frequency of 3484 update not really that different to general
  configuration data update, usually via dhcp)

- noted managed (dhcpv6) networks tend to have managed policy
  (thus dhcpv6 option may be appropriate there)

- noted unmanaged networks probably don't have a policy table available,
  so should explore routing hints etc there?

The draft is on the agenda for the 6man session in Stockholm.

Obviously feedback is welcome at any time.

Tim