Address markets and the transition to IPv6

Some critics argue that a transfer market would slow down or harm the transition to IPv6. A transfer market, they say, might encourage organizations to consider purchasing more IPv4 addresses instead of firmly committing themselves to an IPv6 migration strategy.

Note that this argument implicitly concedes that a transfer market would work. Transfer markets could make organizations defer their IPv6 migration plans only if the transfers succeeded in facilitating reclamation and use of a significant stock of unused IPv4 addresses. In effect, those who make this argument are saying that we should maintain artificial limits on the availability of IPv4 addresses in order to force organizations and ISPs to migrate to IPv6 more quickly.

That policy seems like a dangerous one. The depletion of the free pool of IPv4 addresses could trigger a coordinated movement to IPv6 only if IPv6 was backwards compatible. But it is not. We know that IPv4 addresses will be needed for some time to maintain compatibility between the two Internets as the migration takes place. Unfortunately, many of the compatibility issues associated with large-scale dual stack implementation and NAT-PT are not well understood. The transition could turn out to be more complicated, costly and difficult than anticipated, and we don’t know how long it will last. If we try to use an address shortage to force ISPs into making the transition before they are ready, we could develop damaging gaps in connectivity due to shortages of address resources and compatibility problems.

There are really only two possibilities: either 1) address scarcity in the IPv4 space makes migration to IPv6 inevitable, or 2) migration to IPv6 is not inevitable, and Network Address Translators and other workarounds can keep IPv4 alive indefinitely.

If a migration to IPv6 is inevitable, a transfer market could only prolong the transition slightly, it could not stop it. A transfer market can only shift address resources from organizations who are not using them to active use, and from lower-valued uses to higher-valued uses. Transfers do not create more addresses. Given the uncertainties surrounding the costs, duration, and technical problems of the transition, extending the time horizon for decisions to migrate to IPv6 seems like a good idea. It is certainly a less risky strategy than creating an abrupt shortage.

If migration to IPv6 is not inevitable then any attempt to force people into IPv6 by preventing more efficient use of remaining IPv4 addresses could simply fail, and even backfire badly. Organizations might respond to the artificial shortage with an intensified NAT strategy, or find ways to acquire the IPv4 address resources they needed via subterfuge and underground transactions.

Ultimately, the real impetus to switch to IPv6 is that it can deliver applications and capabilities that IPv4 can’t – including the ability to do things that require larger amounts of address space. When the special applications and capabilities of IPv6 reach critical mass and organizations can realize network externalities by switching to it, a bandwagon effect will start and a complete migration will take place. Since no one can predict when, or even whether, that will happen, we must prioritize efficient and flexible management of the IPv4 address pool. We should not gamble with the fate of the Internet we have in order to push people into one that doesn’t exist yet.

One comment