In your letter dated Thu, 23 Sep 2010 07:33:24 +0200 (CEST) you wrote: >I'm talking about ETTH, one port in an L2 switch is a household. I know >what port goes to each household, so "trust" is not the issue. > >In IPv4 I hand out an IP address and I know to what port (option 82) this >IP address is at, and the L2 environment makes sure this port can only >source traffic from the IP it has been handed for the duration of the >lease.
I wonder. I don't know if this has been discussed or proposed before. Why can't the CPE act as some kind of proxy between the customer's network and the ISP's WAN network? If an ISP essentially wants to run point to point links over what is technically a shared ethernet with some filtering, why not define a protocol for doing that and make sure that that protocol is only between the CPE and the ISP's access concentrator? The CPE can then for example continue sending RS messages, and can terminate NS messages by always returning the concentrator's MAC address, etc. That way, knowledge of the WAN link isolated in the CPE, instead of spread across all hosts and consumer routers. -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
