Re: [dhcwg] What sorts of services does DHCP configure?

[Ole Troan <[e-mail]>]

Ted,

>> Taking a more specific example, "Distributing Address Selection Policy using DHCPv6", draft-ietf-6man-addr-select-opt. For now, this draft only propagates generic address selection policy. It is little controversy to be DHCPv6 configuration, in my eyes.
> 
> It's pretty clear to me that this option is not a very good idea in most cases.   I don't think the DHC working group is in a position to say no to it, but we are certainly in a position to say "this isn't a good general solution to the source address selection problem."

are you sure you don't mean to say, "building networks this way is not a very good idea in most cases"?
because if you suspend enough disbelief, then the requirement for the network to be able to give hosts
information about the SAS/DAS policy table is hard to get around.

a good general solution to the source address selection problem? we don't have one, and I think it
unlikely that we will ever get one. active probing, aka multi-prefix Happy Eyeballs is the only thing that may work everywhere, and even that as a set of problems associated with it.

> Of course, there are constrained environments where it may make sense—e.g., in a PE router<->CE router situation.  But this is not something that ought to be enabled on all DHCP clients.

to solve the use case described in:
http://tools.ietf.org/html/draft-ietf-v6ops-ipv6-multihoming-without-ipv6nat-05

then that does exactly that, recommend it to be enabled on all DHCP clients.

cheers,
Ole