Don't be alarmed by IPv6's declining numbers

Before a tsunami hits shore the sea recedes abruptly and there is a deceiving period of calm before the waves arrive. Given IPv6 address allocation trends, perhaps the analogy applies to the deployment of IPv6.

2006 was a year of progress for IPv6. The 2005 U.S. government mandate - requiring its agencies to move to IPv6 with some level of capability implemented by June 30, 2008 - spurred activity not only in the government but also in the commercial sector that serves the government. International activity, already somewhat ahead of U.S. development, continued to make strides with several Far East countries committing resources and providing government-backed incentives for IPv6 deployment.

But despite what appears to be increased momentum, the number of IPv6 address allocations made by the Regional Internet Registries (RIR) has consistently declined since the peak in 2002.

According to statistics obtained from the Web site of RIPE NCC, one of the RIRs, address allocations peaked in 2002 at 384. Allocations since have steadily declined, with 2006 being the slowest period since 2002, with fewer than 200 allocations.

Does the decline represent a trend, perhaps a decrease in overall interest in IPv6? Well, there may be other factors involved.

Before 2002, allocations were largely for experimental or development purposes. Following the June 2002 release of the IPv6 Address Allocation and Assignment Policy, which effectively opened the door to address acquisition, there was an immediate spike in allocations.

Click to see: IPv6 decline in numbers

The spike most likely was driven by early adopters, organizations actively involved in IPv6 development since its inception, or those ahead of the curve in terms of deploying IPv6 in enterprise networks and commercial services. The subsequent decline likely represents a leveling off of that involvement.

What is certain is that the vast majority of those who will deploy IPv6 eventually have not yet acquired addresses. One reason is that there is little motivation to do so today. And the virtually infinite IPv6 address space eliminates the pressure to rush out and acquire addresses before they're needed. What's more, current policy limits allocations to service providers who will make re-assignments, requiring users to obtain addresses from their upstream providers. This reduces the number of RIR allocations.

In IPv4, strict allocation policies were set to thwart the rapid depletion of address space. Although this policy is critical, it has led to service providers acquiring multiple allocations. With IPv6, initial allocation should be enough to satisfy even large service providers for quite some time.

While the number of allocations has declined, the amount of address space allocated shows growth. There was a drop in 2006, but that is attributable to a few huge /19 and /20 allocations made in 2004 and 2005. Despite the drop, 2006 was still well above 2002 and 2003. Overall, there appears to be growth.

The 2002 policy's standardization of allocation size also is a factor. The /32 standard was defined to provide consistency and structure as well as to maximize prefix aggregation benefits. It also provides enough address space to satisfy many organizations indefinitely, minimizing the number of subsequent requests. Of course, larger allocations can be obtained with proper justification. Although larger allocations have been made, they don't indicate a trend in terms of the number of allocations. They do, however, indicate significant requirements and desire for IPv6.