Very good point Erik, thank you. We will certainly take that into consideration going forward. This is also why we are proposing a mix of two to three CER/ISP Edge determining characteristics (e.g. Including the CER_ID option as well as the /48 prefix check, etc.).
Cheers, ~Chris On 2/25/13 8:08 PM, "Erik Kline" <[email protected]> wrote: >WRT to Section 4.1, I think the "/48 check" may not be a good idea. >It may work for /your/ deployment models, but it doesn't necessarily >work for all. Furthermore, if this were widely adopted it would >effectively enforce that deployment model all providers, needlessly I >think. > >Consider a deployment model that I believe I first heard from Lorenzo: >a customer is allocated a prefix AND the "zero-numbered" /64 from that >prefix is advertised in the RA toward the customers home gateway. >(Really any <64 customer prefix and any single /64 would work, like >the "all ones" /64--it doesn't particularly matter.) I'm not aware of >anything that technically precludes this nor do I think we should move >to preclude it now. > >(Note: there may be some home gateways that are not be smart enough to >recognize that one of the prefixes delegated to it is in use already >on one of its interfaces (e.g. WAN), but homenet can require such >intelligence.) > >So, your text could try to recognize that IF there is only one >allocated prefix in the home && IF an interface is not number from >that prefix THEN it must be an edge. That's fine. But being numbered >from within the allocated prefix does not necessarily mean that it's >an interior router. _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
