Editor's Note: This is the first in a series of posts that will soon be released as a complete IGP paper.

What happens when the Internet addresses run out? That question has been generating growing concern among Internet operators and policy analysts. The problem is fundamental to the future of the Internet. The Internet we know is based upon a data communication protocol known as Internet Protocol version 4 (IPv4). IPv4 is a software procedure that moves data packets from one unique numerical address to the other. The 1981 standard that defined Internet Protocol created a fixed address field of 32 bits, which creates a mathematical possibility of about 4 billion unique addresses.

Because of continuing Internet growth, IPv4 addresses have become scarce and valuable resources. Put simply, we are running out of them. At the middle of 2008, the last remaining stash of unused address blocks – the so-called “unallocated address number pool” – had dwindled to only 39 blocks. In recent times, these blocks of addresses have been distributed to regional address management entities at a rate of about 12 per year, which means we have only about a three year supply left. Demand may accelerate as exhaustion of this pool approaches, but even if it doesn’t the end of the unallocated pool is within sight. The problem of address scarcity is as severe for the Internet economy as the oil shocks and gasoline shortages of the 1970s were to the industrial economy. Address shortages could act as a brake on the growth of the Internet. Emerging Internet economies in Africa, Asia and Latin America are just beginning to fulfill their potential level of Internet development.

In principle, a new Internet standard, IPv6, solves the problem of address scarcity because it has a very large address space (2^128 addresses). But the new Internet standard is not compatible with the old one. Thus, anyone who implements IPv6 and throws away their IPv4 capabilities is going to be isolated, cut off from the vast majority of communication partners, web sites and services associated with the old Internet. One must think of the transition from IPv4 to IPv6 as the (possibly temporary) co-existence of two distinct “Internets.” The only way to keep the two Internets universally interconnected is for adopters of the new standard to run both versions of Internet protocol. This is known as running “dual stacks.” The need for dual stacks means that even IPv6 users will continue to need IPv4 addresses during the transition period. During the migration to IPv6, users will not engage in simple substitution of IPv6 addresses for IPv4 addresses. IPv4 address resources will continue to be in demand.

How long is that transition period likely to be? No one knows for sure, but efforts to model this problem do not yield encouraging results. One group of academics applied mathematical diffusion models to IPv6 adoption and projected that the overlap period under the most favorable assumptions could take no less than 6 years, and might take up to 70 years. There are still Internet technical people who openly claim that IPv6 will never succeed in reaching the critical mass needed to replace the IPv4 Internet. Another recent economic assessment of the incentives to migrate also came up with pessimistic conclusions. Thus, it would be unwise for IP addressing policy to be based on the assumption that a global migration to IPv6 is inevitable. Sound address management should be designed to conserve and reclaim IPv4 resources and ensure that they are assigned to their most important, highly valued uses.

What should be done? This blog post is part of a paper that evaluates a transitional policy that Internet governance agencies are considering as a response to the depletion of the IPv4 address space. In particular, it focuses on proposals to allow organizations holding IPv4 addresses to sell address blocks to other organizations willing to buy them. IP address transfer markets, as they are called, have been proposed as a pragmatic way to extend the life of the legacy IP address space. One important benefit of such a policy is to provide incentives for existing holders of addresses to release unused address resources. Another possible benefit is the way it might rationalize and make more transparent an underground economy in address resources. Transfer markets also increase the autonomy of Internet users by providing an alternative to the centralized administrative processes that currently control address allocations. Fairly liberal address transfer market proposals are being considered in the European and Asia-Pacific regions. A more restrictive transfer proposal is also under consideration in North America.

This paper analyzes the economics of the proposed transfer policies, and conducts a systematic comparison of the policies proposed in the three main world Internet regions. It concludes that:

• Address transfer markets offer a pragmatic solution to the problem of reclaiming a substantial amount of unused IP address space and of re-allocating addresses to their efficient uses

• The risks of instituting well-designed address transfer policies are small when compared to the potential benefits. The change is less radical than it appears.

• A failure to legitimize address transfer markets would create substantial risks of the institutionalization of gray or black markets in IPv4 address resources, leading to a deterioration of accurate registration and administration of the legacy address space. This could have severe negative implications for Internet security.

• A transfer market will not impede v6 adoption but it will give it a longer time to play out. We do not know how long or even whether the global Internet will succeed migrating to IPv6. It would therefore be irresponsible to base IPv4 management policies on an assumption that such a migration will take place.

• The proposed address transfer policies being considered by RIPE and APNIC are more liberal than ARIN’s. All three could be improved in various ways. Most of the legacy IPv4 address space is in North America; thus, the policies ARIN adopts have the most importance and should be formulated with the good of the global Internet in mind. RIPE, ARIN and APNIC should strive to harmonize their transfer policies and make inter-regional transfers possible.

The significance of this issue goes beyond a narrow assessment of transfer markets as a policy. The creation of address transfer policies both reflects and reinforces the growing importance of the Regional Internet Address Registries in global Internet governance. Thus, this paper supports a broader conclusion which will be explored in greater detail in a future IGP paper:

• For transfer markets to work properly, the registration and enforcement functions of the RIRs will need to be strengthened. As the address allocation process becomes more institutionalized, we will need stronger public policy frameworks for RIRs to operate within. These policy frameworks should retain and respect the RIRs’ status as independent self-regulatory entities, but also ensure that their policies are constrained by basic human rights protections regarding freedom of expression, privacy and due process, as well as economic policies regarding competition.

Next: An Analysis of the IP Address Governance Regime