[ppml] Policy Proposal 2005-1: Provider-independent IPv6 Assignments for End Sites - Last Call

[vixie at isc.org]

cathy wrote:

> Thanks for sending this Vince.  I too remember those days when things
> were melting down.   I often feel that no one else remembers.   It is
> hard to watch the same mistakes being made over and over again.

i remember the memory limits of the AGS+ and the original RP.  but these are
different mistakes than those.  i know folks who applied (pre-CIDR) for a
"class C" address for their houses, and got them.  are personal dwellings
likely to multihome?  well, actually, yes.  mine did, back in the day, and
many will, if you consider switching providers without renumbering to be a
derivative form of multihomage.

these are different mistakes than those.  the ipv4 swamp was mostly populated
pre-cidr, and there are still a lot of SMB's and SOHO's therein.  the proposed
ipv6 swamp can be much smaller, and we should constrain it with all kinds of
rules about multiple locations, entity-size, requirements for multihoming,
total allocations per RIR per year.

since ietf did in fact decline to consider multihoming as a first order
problem to be solved during the ipv4->ipv6 transition, and since it's rather
late to resurrect DNS's A6 and DNAME RRs, and since shim6 and mobile-ip and
et-al appear to be hard sells to the gray-or-balding sector of the operator
community, we're left with PI.  but it's not the same mistake as the V4 swamp.
it's a different mistake, owing its heritage to a whole line of other mistakes,
and mistaken or not it's the only proposal i've heard that's got any legs.