Hi Washam, Please see below.
Le 21 août 2011 à 08:01, Washam Fan a écrit : > Hi Remi, > > Please see inline. > >>> 7. in example a in section 6, how come a mapping rule responding to 2 >>> cpe ipv6 prefix length? cpe will get confused when it did forwarding. >> >> CPE prefix lengths may be different. (It is the Domain prefix length that is >> given in the rule.) >> This is key to be able to assign port sets of different sizes (including >> 64K) to different customers (at least when the CPE-cascade option isn't >> necessary). > > Yes. But my concern was, for example, a CPE has the below mapping > rule, cited from the draft: > +--------------------+--------------------+-------------------------+ > | Domain IPv4 prefix | Domain IPv6 prefix | AFTR IPv6 subnet (e.g.) | > +--------------------+--------------------+-------------------------+ > | 192.32../12 | 2001:db0::/28 | 2001:db0:aaaa:aaaa::/64 | > +--------------------+--------------------+-------------------------+ > > when the CPE forwards a IPv4 packet whose destincation address is > within 192.32../12, how the CPE know it will use /48 CE prefix or /52 > CE prefix? Or the mapping rule implies the CPE should use the same CE > preifx length as its own delegated prefix? Ah, I think I better understand your question now. The SRC CPE _doesn't need to know_ the length of the DST CPE IPv6 prefix length. If this DST length is 48, the DST IPv6 /64 can contain port-derived bits that have no effect on the route toward this CPE. (It is reachable at its /48, irrespective of bits beyond the /48). The DST CPE simply makes NO USE of these extra bits (except possibly to check their consistency with the L4 payload). Does this answer you question? In any case, I believe that a later version of the draft should contain an explanation of that kind, preferably more detailed. I will work on that. Thanks for your question. RD > > THanks, > washam _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
