On Apr 15, 2013, at 3:40 AM, <[email protected]> wrote:
> Except "IPv4 addressing"-related configuration, what additional configuration 
> parameters can be provisioned with DHCPv4 and not with DHCPv6? 
> Why in the future, an IPv6-enabled node will require to retrieve 
> configuration using DHCPv4-over-DHCPv6 and not directly using DHCPv6?

That's exactly the point, Med.   We are attempting to separate out legacy 
solutions that continue to require native IPv4 from those that can survive 
entirely on IPv6, which we expect in the long run to be the vast majority.   
The point is simply to say that if you want to configure your IPv4 stack with 
DHCP, you should, in general, continue to use DHCPv4.

There is general agreement that it's pretty harmless, and also quite efficient, 
to provision MAP-E-based solutions using DHCPv6, as long as all they need from 
the DHCP server is their MAP prefix/port set configuration.   But we are trying 
to keep the camel's nose from coming any further under the tent by explicitly 
saying how you do any additional provisioning of the IPv4 stack, should you 
need to do so.

This document also addresses the problem of doing _stateful_, as opposed to 
_stateless_ IPv4 address configuration for dual-stack nodes on an IPv6-only 
network with IPv4 tunneling to AFTRs.   In that case, again, the required 
extensions to DHCPv6 would be large, and it's actually pretty convenient to 
just leverage the DHCPv6 infrastructure to carry protocol messages to a DHCPv4 
server, which already knows how to do dynamic IPv4 address allocation.

_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to