The False Scarcity of RIR IPv4 Exhaustion

An efficient global IPv4 market will provide the Internet community 5 to 7 more years on IPv4 to facilitate a migration to IPv6 that follows normal technology adoption patterns.

In our last blog post, we wrote that depletion of the RIR free pool of IPv4 numbers isn’t the end of the road for IPv4.   Two recent articles written in Information Week discuss the meaning of IPv4 exhaustion from different and opposing perspectives.

On November 10, 2014, Lawrence Garvin wrote a piece entitled “The IPv4 Sky Really Isn’t Falling.” A week later, ARIN President and CEO, John Curran, responded with his own article on the subject, “Why We Must Move to IPv6.”  These articles represent two common perspectives on the IPv4/IPv6 “conundrum” and both make some valid points.  They also omit critical information that weakens their respective arguments.

Garvin argues that IPv4 numbers are not actually exhausting. The “shortage,” in his view, is really the consequence of number allocation mismanagement -- not real scarcity.  He correctly observes that nearly all IPv4 numbers have been allocated but there are, in his estimate, at least 400 million unused previously allocated IPv4 numbers in the U.S. alone.   He then, without regard to the commercial and legal implications, states that companies should give their unused (or, more accurately, unadvertised) numbers away for free to ISPs, CDN and others to help them grow their IP networks (and revenues).

In his responsive article, John Curran argues that “the reality” is that ARIN will deplete its remaining pool soon.  When that happens, Curran asserts, no more numbers will be available in North America, which he believes explains why IPv6 deployment is well under way and why everyone should migrate now.  Curran acknowledges that there is some previously allocated IP address space that is not being used but speculates that this supply isn’t enough to satisfy 3 years of IPv4 demand.   Curran, like Garvin, notes that DOD and a handful of others have returned addresses and that there are policies in place to encourage others to do the same.

Both Curran and Garvin fail to properly account for the growing IPv4 market.   The active, but early stage, IPv4 market is successfully putting unused IPv4 numbers in the hands of those companies that need them.   When ARIN reaches depletion of its free pool, ISPs and others that need additional IPv4 numbers can purchase them from private sellers in the competitive marketplace.  The 15 /8 blocks referenced by Curran (approximately 250 million numbers) and Garvin’s 400 million available numbers in the United States both likely significantly underestimate the supply of numbers that may be available for reallocation in the private market.

There are approximately 1 billion unadvertised previously allocated numbers globally.   Add to that some currently advertised IPv4 address space that could be renumbered to free up underutilized blocks for sale when an efficient and transparent market offers existing holders enough financial incentive to incur the renumbering expense .  As Curran observes, the annual global average RIR allocation rate between 2000 and 2010 was approximately 160 million.  Growth of end points on the Internet is not linear – the growth in connected devices in 2010 was certainly greater than the growth in 2000.  But increased usage of NATs and CGNs over the last several years has dampened the actual consumption rate of globally unique IP addresses.   With this historical allocation rate and available supply of previously allocated numbers, a global IPv4 market could satisfy growth of the Internet for approximately 6 years after complete RIR free-pool depletion.

Market-based re-distribution of IPv4 numbers is certainly a transitional solution.  After real IPv4 exhaustion, the transition to IPv6 must be largely complete.   An efficient, transparent and global IPv4 market will provide the Internet community 5 to 7 more years on IPv4 to facilitate a sensible transition following the natural pattern of technology adoption.