On Nov 9, 2003, at 1:24 PM, Brian E Carpenter wrote:
Alain,
Please define "real PI (by real I mean registered)". Not having seen the
draft that defines it, I can't evaluate your argument.
The problem with the Hinden/harbeman draft is that it allocates a part of
the public IPv6 address space for private, unregistered usage.
The consequence is that when (and not if) those addresses
will leak in the public Internet, they will be untraceable and won't resolve at all
in the reverse tree DNS.
My suggestion is to let the authority in charge of administering
the public IP address space to allocate directly address space
from a specific bloc to whoever wants it, with no expectation that
it will be routable and leave it up to the customers and their ISP
to see if this gets routed or not (with a recommendation that by default it is not).
That way, those addresses would be traceable the day they will leak. Of course, this requires a little extra management, and perhaps a (small) recurring fee to maintain the database instead of a one time 10 euro, but this should not stop anybody serious.
If you, or the wg, thinks this avenue is worth exploring, I can write a 2 page draft. I honestly believe that this entire issue can be solved outside of the IETF by the RIRs without introducing anything new/damaging in the IPv6 architecture.
- Alain.
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
