In 2010, we introduced IPv6 RIPEness to recognise Local Internet Registries (LIRs) in the RIPE NCC service region that had begun deploying IPv6. Seven years later, we looked at how the IPv6 RIPEness level of the RIPE NCC membership evolved over time.
The IPv6 RIPEness system awards LIRs with up to five stars for each measurable step towards IPv6 deployment. As described in this article, every LIR can reach four-star RIPEness by requesting IPv6 space, registering a route object, requesting reverse DNS and announcing the IPv6 space in BGP. A fifth star was introduced in 2013 to measure actual IPv6 deployments.
In previous articles we looked at the present state and compared how the LIRs of different age groups were doing. Now we look at it from another angle: how did the IPv6 RIPEness level of the RIPE NCC membership as a whole evolve as a function of time? Because the sampling methodology of the fifth star makes this more difficult to reach for smaller LIRs, we restricted the analysis to the first four stars, so that each LIR has equal opportunities to be included and counted. Reaching four-star IPv6 RIPEness is entirely in an LIR's own hands.
Figure 1: IPv6 RIPEness levels of all LIRs as a function of time. The vertical red lines represent the times when, respectively, IANA ran out of IPv4, RIPE NCC reached its last /8 of IPv4 and the IPv6 requirement was dropped from the last /8 policy.
In Figure 1, we plot the number of LIRs with zero, one, two, three and four stars of IPv6 RIPEness respectively for each day since we started with IPv6 RIPEness. We also have monthly data points for the years 2008 and 2009 as well as the first half of 2010; these were computed retroactively from archived data.
The evolution of IPv6 RIPEness shown in Figure 1 is characterised by four time periods, marked and delineated by the vertical red lines:
- August 2010 - July 2011. Running up to and immediately following the day IANA handed out the last of the unallocated IPv4 /8s to the RIRs (3 February 2011), the number of four-star LIRs was rising rapidly. Members who realised it was time to act on IPv6 took the steps that brought them four-star IPv6 RIPEness.
- July 2011 - September 2012. After the initial surge, growth in all IPv6 RIPEness categories slowed down in the second half of 2011. Growth continued at these levels, with the four-star category being the most populated one, until September 2012.
- September 2012 - March 2015. The day the RIPE NCC's IPv4 pool (first) fell to the level of one /8 marks a turning point in the evolution of IPv6 RIPEness. In an effort to promote IPv6, stimulate deployment by LIRs, the IPv4 policy for allocations from the last /8 required LIRs to hold an IPv6 allocation, either from the RIPE NCC or from another LIR, before they could get the final /22 of IPv4 space. Because of this requirement, we see a sharp increase in the growth rate of one-star IPv6 RIPEness. Both new and existing LIRs who were in need of IPv4, now had to obtain IPv6 space and thus scored the first star of IPv6 RIPEness. Two-star IPv6 RIPEness picked up growth as well, but to a much lesser extent. Three- and four-star IPv6 RIPEenss, however, continued growing at the same rate as before the RIPE NCC reached exhaustion.
- March 2015 - present. On 4 March 2015, the RIPE community reached consensus on the policy proposal which removed the IPv6 requirement for receiving space from the final /8. Implemented by the RIPE NCC two weeks later, this marks another turn in the evolution of IPv6 RIPEness. With the requirement gone, not all new LIRs are requesting IPv6 address space from the RIPE NCC. The number of zero-star IPv6 RIPEness LIRs started to increase again, and growth in one-star IPv6 RIPEness slowed down. Two-, three- and four-star RIPEness on the other hand continue to grow at the rates set at or before the time the RIPE NCC reached the last /8, about four and a half years ago.
Figure 2: Relative levels of IPv6 RIPEness for all active LIRs
Figure 2 shows the evolution of IPv6 RIPEness levels as a percentage of the total number of active LIRs. The growth in one-star IPv6 RIPEness LIRs between 14 September 2012 and 19 March 2015, when holding an IPv6 allocation was required in order to receive the final /22 of IPv4, is again quite visible. The percentage of LIRs with four-star IPv6 RIPEness did not change significantly during that time. It rose from 18.5% on 14 September 2012 to 21.5% on 14 December 2014 and has hardly changed since. The percentage of LIRs not holding any IPv6 space stopped decreasing after March 2015, leaving a more or less constant 75% of all LIRs with one or more stars of IPv6 RIPEness.
The conclusion we can draw from these observations is that the policy of forcing LIRs to obtain an IPv6 allocation did not achieve the goal of making LIRs more IPv6 ready. The largest increase in IPv6 RIPEness was seen in the one-star category; i.e. the star which LIRs got automatically when requesting the mandatory IPv6 allocation. Three- and four-star IPv6 RIPEness were growing at about the same rates before, during and after the requirement was present in the policy. This shows that those who want to move towards IPv6 deployment only need encouragement, not a hard policy requirement.