[arin-ppml] Mighten it happen like this?

[Ted Mittelstaedt]

In thinking about this IPv4 runout it occurs to me that it might possible
go something like this.

After the last /8 is assigned to ARIN, the hostmasters there will get out
their knives and start slicing off subnets from it.  A handful of Very Large
requests will be satisfied from it, then a lot more smaller requests, then
the /8 will be as the turkey is on the platter during the last Thanksgiving
-
the large breasts gutted but still plenty of edible meat in smaller chunks.

Then the reclamation efforts will start turning up even smaller and smaller
chunks of IPv4 abandonded years earlier - nothing to satisfy the large
consumers,
but still plenty of smaller tasty bits.  And in the meantime they will still
be stripping the carcass of the /8 for the last usable bits.

Then the carcass will be done and reclamation will be in full swing now -
but
the IPv4 coming in from reclamation will be somewhat less tasty -
ex-spammers
blocks, stinky old swampland that may or may not have been used.

Then reclamation will start petering off and the IPv4 bits coming in from it
will be very nasty indeed - blocks with squatters in it that the obtainer
of the block will have to actively evict, blocks where the original occupier
is still fighting with ARIN over ownership.

Then they will get down to the nitty-gritty of trying to piece together
minimal
sized blocks to allocate from scattered /24's some of which are abandonded,
some
not - begging and pleading with owners to please move over to this other /24
so
we can use the one your on to aggregate together a larger block.

Somewhere along the way some kind of transfer market may spring up - short
lived
though it may be, with a few folks making a killing off selling blocks - but
as
time passes it will die down.

During this time the number of orgs wanting IPv4 will be decreasing as more
and more of the requestors give up hope of getting usable IPv4 and more and
more
of them migrate to IPv6.

So, perhaps maybe fully 4 years after the last /8 is allocated to ARIN then
the
very last aggregated subnet of any size will be given out - reclamation will
be
exhausted, and pretty much nobody will have any hope left of getting IPv4
allocations
except from the transfer market.  That might mark the "official" end of
ARIN-assigned
"free" IPv4.

The transfer market will be peaking right around now - as prices get so
rediculous
that it becomes cost effective for even the most retrogade networks to go to
IPv6.

Then a tipping point will be reached and over a few months the bottom will
drop
out of the transfer market and the market will crash, and we will see the
commencement
of an accellerated schedule of more and more networks dropping IPv4.

A few years after that then reports will begin to show up of routing
unreliability
of the IPv4 Internet in certain spots on the Internet.

By 4 years after the "official end" of ARIN-assigned IPv4, we will start to
see
websites set up with countdown clocks predicting the very last day that IPv4
traffic
will exist on the Internet.

Then, sometime in 2020, some politician will send an e-mail titled "Goodbye"
at a
ribbon-cutting ceremony that will mark the last time that a real IPv4 packet
will
be sent over the public Internet from a public client to a public server.

Around 2025, Cisco will make proficiency in IPv4 an optional part of it's
assesment test.

Around 2030, Juniper and Cisco will release firmware that won't have IPv4
support in
the base load.

Does seem like a realistic end-game?

Ted