[cc trimmed] On Tue, 2006-03-28 at 01:54 -0800, Michel Py wrote: > > People will still want to do NAT on IPv6. > > Yes, and since site-locals have been deprecated they will also hijack an > unallocated block of addresses to use as private, same what happened > prior to RFC 1597 for the very same reasons (difficult/pricey to get > PI). I guess you missed out on: http://www.iana.org/assignments/ipv6-address-space FC00::/7 Unique Local Unicast [RFC4193] You can use that to generate your local prefix and it is much better than site-locals as the chance of collisions is fairly low. And as you know you simply don't want to do a NAT from 10/8 to 10/8 at one point in time when two big companies merge ;) > >> Michel Py wrote: > >> A protocol that would be only v4 with more bits in the first place, > >> with routers / NAT boxes that would pad/unpad extra zeroes (also > >> including extra TBD fields). As this would be 100% compatible with v4 > > >> this could be deployed without too many headaches. (I almost got lost in the attribution level here) Then why is IPv6 causing so many headaches? As one can see 6to4 is mostly making up your IPv4+ address from the IPv4 one by doing: 2002 + <ipv4 address> + padding bytes ;) Ah, of course, one actually need to upgrade most of the internal stuff and upgrade all the applications, convince managers, get money to do it, do training.... etc... Also for the rest of the thread, overlaying IPv6 over IPv4: RFC4380 Which is more or less a p2p overlay network using IPv6 as the addressing part and thus leveraging a lot of applications already. Greets, Jeroen
Attachment:
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf