2011 is the year that IANA will exahust the central pool of IPv4 addresses.

This page links to a collection of graphs related to the IPv4 pool.
These are snapshots in time related to past behavior. As such, any
projections should be considered “best case” because future
mob behavior is likely to result in a run-on-the-bank.

The state of the entire IPv4 pool as allocated by IANA:
IANA pool

The IANA pool burn rate flattened after this was originally
shown in mid-2005. This graph shows a smoothed version of the
actual intermittent allocations. At the point where the RIR's
have exhausted their internal pools, this graph will have
to return to a growth pattern that matches the actual outbound
compound growth rate from the RIR's.
IANA burn rate

The next graph shows the annual distribution rate for each RIR
to their customers. The current year includes a year-to-date based
projection as a simple daily normalized rate multiplied by 365.25.

Note that the APnic rate continues unabated, around 3/4 of a /8 per month.

It does not attempt to adjust for annual rate differences due to
holiday periods. One thing that it does show is the global economic
impact, as the 2008/9 rates look very similar to the 2001/2 rates,
and that the first half of 2010 is dramatically up.
RIR delegations to customers

Taken together, the previous two graphs show that the draw rate
by the RIRs does not match the outbound rate to their customers. To
accomplish this, their internal pools have been diminishing. Policy
allows for each RIR to have an internal pool sufficient for 2 years of
projected need. While this graph is based on actual past allocations to
customers, the only way it could represent 2 years of projected need is
for each RIR to expect a decline in demand. The result means there will
not be an extended period that many expect between the exhaustion of
the central IANA pool and the individual RIR pools.
RIR pool depth trend

Basing the projection from the inflection point in the IANA burn
rate at the year 2000, the end of that pool keeps shifting in time
due to the intermittent nature of the IANA to RIR allocations.
At the same time, this is the focus of most people’s attention.
IANA exhaustion date

Basing the IANA projection on just the recent data shows a much
different result. Essentially the current trend projects exhaustion
about 9 months sooner than the longer term trend. Many would consider
this to be expected behavior as we near the end of the pool.
IANA exhaustion - recent trend ...

A much more stable graph is the one based on the total pool
including the RIR internal holdings with the allocations to end
customers. This one has not changed significantly since 2005, though it
shows the collective RIR set rather than individual holdings.
RIR exhaustion date

The combined view is inverted because I am too lazy to fix the tool
but it shows the trend of each RIR, along with their collective trend,
and the central IANA pool. At various points in time the top 2 curves
have crossed, as the RIR draw rate has dipped well below their outbound
rate. This is clearly unsustainable and can’t happen in practice.
The white lines show the time window between IANA to RIR cumulative
allocation, vs. the allocation of that many from RIRs to end customers.
Historically this was intended to be 24 months, but as the end nears is
trending toward 6.
Combined view

None of those predictions are particularly accurate, because they
include the last 5 /8's as 'available', while IANA and RIR policy have
already set those aside to be allocated for transition to IPv6 uses.
Taking the smoothed IANA burn rate at ~ 1 /8 per month, those 5 final
/8's represent around 5 months shortening in the IANA curve. It is
harder to calculate the impact on the collective RIR pool as their
run rates are all different, so as each drops out of the pool that
date will suddenly make dramatic changes.

So we are at the effective end of the pool. All that remains is for
APnic to come back and pick up their last pair, then IANA will distribute
the remaining 5, one to each for transition allocations.
As of their January 21, 2011 assignments, APnic is sitting about 1/2 of a
/8 below their recent low-water mark, and would have already requested more
from IANA in normal times. Their internal pool will carry them to around
mid-March, and given that there is little chance anyone else will qualify for more space in that
timeframe there is little reason for them to rush it to defend against
others potentially taking those last 2. There has been talk about a well
orchestrated global press release, so given we are already a week overdue
I would expect the text for that to be circulating between IANA and the RIRs
now. When it will be released is the open question.

The next event to watch is the date when an RIR has to turn members away.
Given their much higher allocation rate, currently it would appear that APnic
will be the first to do so, and as of Jan. 24, 2011 it appears they will do that
before June 1, 2011, likely early in May. This date is highly dynamic as it is
based on the most recent allocation data, and past rather than the real future
allocations.

An interesting thing to note is that during the height of the dot-com bubble
the collective RIR distribution rate to their members was the equivalent of a
/24 every 3 minutes. Now under much more stringent policies about who qualifies
and how efficient the space must be utilized before getting more, for the last 12
month period the event window of a /24 equivalent has been 36 seconds.

A Fractal Map of the space was developed by Randall Munroe.
Sequencing that over time provides some perspective.
Fractal Map

Taking the big picture view, the 30 year graph of the IANA pool is
an interesting thing to contemplate.
30 year

EMAIL: tony@tndh.net ahain@cisco.com

Link: http://www.tndh.net/~tony