I think another point is that if they're concerned about having to run a separate DHCPv6 client "process" to handle PD (as was I think discussed in an earlier email), there's nothing in the DHCPv6 specification that says you can not implement DHCPv6 in the IPv6 kernel. If PD is integral to your network architecture, why not put the implementation there?
You can also have it in both places (though of course likely not the best design if both are doing PD) -- just be sure to allow applications to also use the port and deliver all (or at least those not "accepted" by the kernel layer) to those applications. - Bernie -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
