A robust, but polite and even tempered discussion around transfer and data accuracy at the LACNIC policy sessions on Wednesday, here at the LACNIC 24 meeting in Bogóta, Colombia got to the ‘heart’ (or head) of the problem:
How do you balance the conflicts between maintenance of address management in the public interest through justified need, and the emergence of ‘grey market’ transfers of effective use outside of the RIR registry system?
What is clear from the discussion, and subsequent calls for consensus, is that the LACNIC region membership is not afraid to stand up at the microphone to defend deeply held positions. Several speakers were quite overt in their support of returns to registry of unused resources, and against any acceptance of property rights in the use-management of a block of address.
This is despite ARIN, APNIC and RIPE NCC having negotiated consensus address management policy changes which explicitly recognize the ability to buy and sell addresses, and transfer addresses outside of merger and acquisition processes involving the legal entity behind the address blocks, as well as the emergence of bilateral inter-RIR address transfer policies in these three regions. This is not to say that ‘justified need’ doesn’t exist in these RIR policy processes, but the decision around who has use of a block of address, subject to meeting justified need considerations is now vested in the seller, and can be based solely on price (all other policy processes being respected).
David Hubermann from Microsoft presented what I could call an ‘asymmetric’ address transfer policy seeking to enable the import of address ranges into the LACNIC region from other regions, noting that companies with holdings in other RIR regions being deployed in the LACNIC region might wish to manage them in their own timezone(s) and languages, in region. This proposal failed to achieve consensus quite substantially, and was withdrawn. Opposition from the floor was vocal, direct, but polite. They do not believe this proposal aligns with the values LACNIC address holders wish to promote in the region.
By contrast, a refinement of the existing inside-region transfer policy proposed by Ricardo Patera, which tightened the address holding time from one year to three years, was strongly supported. Several speakers suggested they preferred the original five year holding proposal which had been softened in this policy change. The primary concern here is the feeling that address applicants made statements to use of the addresses, which the wider community had a right to see actually enacted. Transfers inside the time window imply a lack of intent to use the blocks. Some people proposed more active checks on usage and reclaim of unused prefixes, a prospect which (if I can intrude a personal note) I suspect would prove extremely hard for an RIR or NIR to enact.
Mike Burns presented on the data accuracy problem: the observance that functional use of addresses and sub-ranges of holdings marked to another entity clearly exist, and if they are not accurately recorded in WHOIS and other public registry data views, then the utility of the registry system can be called into question should other sources of more accurate address-use data be available. Noting this, speakers remained broadly comfortable with holding the line, and responded mainly to defend restriction of access to registry data services to real address holders was necessary to ‘hold the line’ against transfers by soft policy changes.
LACNIC is not alone in having to confront the clash of culture over Internet address management. Technology like this having grown in the main from public sector, education, health and research investments in most economies (for example, if you consider IPv6 penetration in most economies, it is largely led by the tertiary education sector and research networks, before reaching commercial national-scale deployment) it is inevitable that public interest bodies, and public interest participants see a strong reflection of these goals and aspirations: public benefit. But set against this, national-scale network deployments demand large capital investment, and in the absence of government policy willing to undertake this initiative, it has largely fallen to the private sector, which increasingly recognizes processes it uses in all other areas of its control are not entirely in alignment with public address management process. You can buy all the technology you want in open-market processes, but to source the addresses to deploy on them, you have to engage in a process which is outside these market processes. Whilst in the past this was merely a logistical process to obtain blocks at scale reflecting your planning, it now consists of being permitted to plan for scale, which cannot be met by the process. Instead, you are driven to using NAT, CGN, or to sourcing addresses outside of the allocation process.
In three of the RIR regions, the process is no longer able to supply that demand from held stocks, and has recognized a limited market process as enabling continuance of the growth of technology, while maintaining some level of resource for new entrants, new ideas, and new participants in the network. LACNIC has this vision too, and has done a lot of work modelling the pace of rundown in their phased process of exhaustion of the IPv4 stocks. It is highly likely they will enter ‘phase 3’ exhaustion by the end of this year, when these questions of how to manage public interest and commercial reality will cease to be theoretical, and confront the realities of policy and public interest.
As in other RIR regions, LACNIC staff are professionals, and do not mistake their own views with the requirement to meet the membership consensus policy requirements. Oscar Robles spoke very carefully, narrowly but also calmly to the problem, pointing out that market processes are going to exist whether LACNIC policy process chooses to recognize them or not, and so the discussion is at some levels, moot: It’s not about if markets will exist or not, it’s about what we do about them.
There was a weak consensus outcome with no strong signal of intent, and this question will inevitably return to the mailing lists and public forums for continued discussion.
Speakers universally wanted IPv6 to be the logical outcome. Many suggested that restraint on trading of IPv4 might act as a strong signal to migrate, since the problem space in IPv6 is of a different quality and character. Others felt that markets would perpetuate IPv4 models of Internet forever.
A mixture of head and heart was on display, but no lack of engagement or participation. This was one of the most vocal, involved and committed policy sessions I have attended in a long time, but refreshingly free of heat, and rancour.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.