IANA’s IPv4 pool was officially exhausted in early 2011; Regional Internet Registries (RIRs) are gradually eating through their remaining IPv4 reserves and, although there will always be a trickle of recycled IPv4 addresses coming through as businesses go bust or ISPs move entirely to IPv6, the bulk of RIR IPv4 activity in future will be maintenance of existing allocation records (including associated RPKI certification efforts).
While IPv6 is definitely the way of the future for the Internet, the sheer size of the IPv6 address pool, combined with simplified allocation policies that have deliberately reduced barriers to entry, means there are very few organizations that can’t get IPv6 directly from the RIRs these days.
As a response to these changes, an APNIC policy proposal, prop-103, popped up on 9 July 2012 suggesting that there is no need for any more IP address policy development.
Early reaction to the proposal has been mixed. Some have agreed that it is pointless to play around with IPv4 policies any more; some think that IPv4 policy may still need to be changed. There is greater agreement that IPv6 policy may still need to be refined at some point in the future when there has been more experience in IPv6 deployment.
In a larger Internet governance context, however, the APNIC proposal highlights a major challenge facing the RIRs: what is their role in a post-IPv4 world?
Where will RIRs get their income?
Each RIR has a different fee structure, but broadly speaking the current RIR business model means that the majority of RIR income comes from membership fees associated to IPv4 allocations and management.
Looking forward, ISPs are likely to request far fewer allocations in IPv6 than in IPv4 due to the large size of the initial IPv6 allocations made by RIRs. The smallest allocation size, a /32, is likely to be sufficient for a significant percentage of ISPs, meaning that for RIR fee structures based on the graduated tiers of IP address holdings, most members, if judged solely on their IPv6 holdings, will be in the smallest tier. Currently, fee schedules for IPv6 address allocation and management provide a far smaller portion of RIR income, and given the above, will continue to do so, unless member fees associated with IPv6 are radically increased. Given RIRs are membership-based organizations, there is probably going to be little member support for such an increase.
As RPKI is deployed, RIRs may find it possible to turn certification into a non-insignificant source of income by charging “legacy” IPv4 address holders for certificates that regular RIR members receive for free. “Legacy” address holders, by the way, are organizations that received their IPv4 addresses before the RIRs were created, or before RIRs had member agreements that ensured members had to pay regular address maintenance fees. For a recent article on legacy addresses, see Legacy IP Addresses on Janet.
Unless RIRs make significant hikes in their IPv6 fees, or can make RPKI certification a financial success, however, the reality is that RIRs, in the next few years, may find themselves with significantly smaller incomes.
What will happen to RIR community dynamics?
RIR members and stakeholders have traditionally met and developed their sense of being an IP addressing community via RIR mailing lists and RIR meetings.
RIR meetings began for the purpose of developing policy and, to support the ability of RIR stakeholders to participate in policy making, meetings have included presentations and discussions on industry trends and practices. In a sign of the times, however, APNIC has dropped the use of “Open Policy Meeting” in favour of the policy-free title, “APNIC conference”. AfriNIC and ARIN are now the only RIRs that call their meetings “Public Policy Meetings”, but if the sentiments expressed by Randy Bush in his prop-103 prove true for all regions, it might only be a matter of time before these RIRs change the focus of their meetings, too.
Today, RIR meetings and mailing lists attract a mix of those who want to influence IP address policy development, those who want to understand the policies so they can successfully request resources, and those who are there to build their knowledge on the latest IP address-related industry developments.
With IPv4 becoming less important, both in terms of allocation and policy development, and with IPv6 space large enough and being governed by policies liberal enough to probably need only minor tinkering over time, will the policy-focused community members no longer be attracted to RIR discussions?
With simpler and easier to understand IPv6 policies, will RIRs also find that ISPs don’t need to attend RIR training to figure out how to ask for addresses in future?
Will RIRs find themselves with a community that transforms itself to become ISP staff primarily interested in learning more about industry developments? If so, this leads to the next question:
What is the future core business of RIRs?
RFC 2050 states, “Regional Registries provide registration services as its primary function”. Over time, to support this primary function, RIRs have developed a range of supporting services. However:
- A post-IPv4 world won’t require the current number of trainers, help desk operators or document writers to explain the complexities of multiple-choice policies for IPv4 allocations and assignments.
- A post-IPv4 world doesn’t require as many hostmasters to evaluate IP address requests, not only because IPv6 criteria are simpler and easier to meet, but also because the overwhelming majority of organizatons will never come back with an additional request for addresses. Compare that to the last 10 or so years of RIR operations, where ISPs have regularly returned with additional IPv4 requests.
Will RIRs decide that their core function, IP address management and distribution, has become a lighter function that can be met by slimmed down Secretariats?
A world in transition from IPv4 to IPv6 most probably will require more training and documentation on IPv6 deployment and dual IPv4/IPv6 networks. Will RIRs look to these as their new main activities and sources of income? Or will RIRs refocus their central functions to be RPKI certificate authorities, or expand training departments to cover a wider range of technical training (some of the RIRs already offer training courses outside the strict IP address management function that is the RIRs’ current core business), or reframe RIR meetings and events as regional hubs that provide a platform for Internet operators to exchange best practices?
To place an emphasis on training services would require RIRs to compete with commercial training providers. If this is the case, the probability is that the training costs will need to be borne by those being trained, rather than being subsidized by member income, as is the case today. To place an emphasis on becoming platform for Internet operators to exchange best practices could work nicely, if RIRs work as partners with existing regional Network Operator Groups.
What does the potential change in RIR focus mean for their role in Internet governance?
If the RIRs do reframe their business models, what does this mean for their place in the Internet governance ecosystem? Currently, the five RIRs occupy an impressive three of the 10 places allocated to the academic and technical community on the IGF MAG.
The RIRs, with their traditional focus on technical policy development, have generally been able to occupy the high ground in Internet governance discussions. For example, while domain name policy and governance is seen to be open to influence from all sorts of conflicting interests (intellectual property advocates, privacy advocates, business entrepreneurs, etc.), criticism of addressing policy is less likely to be about conflicts of interest, and more likely to be about interpretation of statistical distribution of addresses (witness the debate in the recently-closed ITU IPv6 Group where some member states came to the conclusion that RIRs were failing to achieve equitable distribution while the RIRs were able to use the same statistics to conclude the opposite.) or whether something is or is not a technical trend that address policy should be changed to accommodate.
If RIRs cease to be organizations that have policy development as part of the core of their new business models, does this also change their position within the Internet governance world? Do RIRs become more like domain name registrars, focused more on resource distribution (albeit without the focus on profit)? Or, if RIRs reposition themselves to be closer in function to the Network Operator Groups (NOGs), do RIRs embrace a more NOG-like role in Internet governance? That is, NOGs seem to play a role in regional and local IGF initiatives, and similar capacity building efforts, but play little role in the larger political Internet governance sphere.
And tangentially, what happens to the ICANN Address Supporting Organization if RIRs change focus?
Could a change in focus give RIRs the ability to survive removal of their IP management functions altogether?
If RIRs downsize and/or change focus, a possible side-effect is that they will lose the capacity to respond to any future attempts by other organizations to relocate IP addressing activities to another form of institution-most likely intergovernmental. (I’m not going to discuss the pros and cons of moving IP address management into an intergovernmental framework here. There has been a lot written about that already, and I’m sure more will be written in future.)
Right now, with RIRs rightly being well-known for their IP address delegation work, the notion of addressing functions being moved elsewhere seems unthinkable. Not only because the current possible alternative homes for such a function are seen as antithetical to the Internet’s multistakeholder, bottom-up approach to governance, but also because to remove IP address functions from the IRs right now would be to remove the beating heart of the organizations. Without that heart, the RIRs would, right now, almost certainly not survive.
However, if the RIRs do, indeed, refocus their central activities outside pure address management, perhaps the RIRs could survive removal of addressing functions and become Regional Internet Registries of Internet technical information and practices, instead of registries of address information. And perhaps become homes for the regional NOG activities (AfNOG, NANOG, etc.).
Time for the community to discuss
Randy Bush’s prop-103 touches on the most immediate change facing the RIRs: the declining role of address policy in the RIRs. The proposal has sparked off some interesting discussion on the APNIC Policy SIG mailing list. But for me, the larger discussion the RIR communities need to have is what they want the RIRs to become in a post-IPv4 world. RIR surveys have touched on this, but as we all know, there are a couple of drawbacks with surveys:
- As I learned well during a short stint working for a PR company specializing in the oil and gas industry, survey questions are often framed in ways that encourage participants to respond in ways the survey authors find favourable to their own goals.
- Surveys tend to attract a very small sample of the stakeholders, and don’t necessarily reflect the true diversity of stakeholder opinions.
Like ICANN, all RIRs have sections of their meetings that encourage the open mic discussions. RIRs have mailing lists for discussion too. I hope RIR communities take the opportunity afforded by Randy Bush’s proposal to starting think about what forms they’d like RIRs to take as they move towards a post-IPv4 world. After all, the RIRs are community-based organizations, and it is through the active participation of the community that the RIRs will be sustained, in whatever form RIRs may take in future.