Tim Chown wrote > > So they can use addresses from the probabilistically unique range under > the space fd00::/8. There is, in terms of raw usage, no difference between > using fd00::/8 or fec0::/10. External networks would still have to route > the prefixes back to you for you to be reachable, which is just as hard/easy > under either system. > Then why not jsut reuse the FEC8:: through FECB:: addresses that were originally the site locals, and are now going to have to be left unallocated for a while? It seems a little wasteful to allocate another range when we have one that is "effectively lost".
> Well, addresses under fd00::/8 and fd00::/8 can be used locally just like > site-local addresses, only they have the nice property of significantly > reduced (probabilisticly unique) or complete removal of (registered unique) > ambiguity, so I don't see where your concern is? > > It is not unlikely that people will be lazy and just use fd00::/48 for sites, > and thus add back in great ambiguity to the probabilisticly unique address > space. First you ask a question, and then answer it. I am concerned that the many network "experts" out there that are trained in a 2 week certification course will take what is taught to them as an example and will make it gospel, and thus use the exact same addresses in multiple networks in close geographic proximity, and thus on the same Carrier edge router. Consider what happens when a carrier implements IPv6 in a city, and suddenly there are 10 companies connected by inexperienced network "experts" (and they have the certificate to prove it) that all follow the exact same course "template". Now this one carrier edge router is connected to 10 incorrectly configured routers all using the exact same "probabilistically unique address". this is where the original need for RFC1918 came from. I still prefer to have this more defined so that the providers can block a range as part of their access lists or filters. Eric -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
