On March 15 and 16 the International Telecommunication Union (ITU) holds a meeting in Geneva focused on internet protocol version 6 addressing policy. One of the focal points of this meeting will be a paper I wrote for the ITU exploring the economics of IPv6 addressing, “Economic Factors in the Allocation of IPv6 Addresses.” The paper put forward the idea of a Transferable Address Block Lease (TABL). This would be a set of address blocks, ranging from /48s to /32s in size, that would be allocated on a provider-independent basis to anyone willing to pay a recurring annual fee based on the size of the block. There would be no “needs assessment,” just a fee.

The TABL proposal was designed as a cautious experiment. The report proposed that only a fixed, relatively small part of the v6 address space would be devoted to TABLs: a /16 should be set aside for this purpose. The rest of the space would be assigned and allocated in the normal way, based on what the RIRs call “needs assessments” (and what political economists would be more likely to call “Soviet-style central planning”). Within this range TABL block leases would be transferable; i.e., in the unlikely event scarcity developed their holders could sell the block as a whole to others or buy more of them in a secondary market. But they could not be de-aggregated for resale in smaller chunks. Of course, portions of them could be subleased or assigned to end users just as ISPs currently do with their assigned blocks.

Since the study was released last fall, it has gotten caught up in the longstanding battle between the Regional Internet Address Registries (RIRs) and the ITU. That is, a few within the RIR camp have criticized the study, not because they have actually read it and have a coherent critique of its proposals, but simply because it was commissioned by the ITU. Similarly, there are those in the pro-ITU camp who have assumed that because the report was commissioned by the ITU, critically assesses the RIRs’ IP address policies and proposes reforms that it somehow favors a greater role for nation-states in IP addressing.

As we explained in an earlier blog, there is an historical rivalry between the RIR/ICANN regime and the ITU, one that frequently manifests itself in debates over control of critical internet resources. Policy debates about the proper economic incentives to apply to the ipv6 address space to encourage appropriate levels of utilization, route aggregation, conservation and reclamation needs to transcend this institutional animosity. What matters to the public is the efficiency and flexibility of ip address allocation. Public dialogue about addressing policy, in other words, needs to focus on actual policies and their effects on users and suppliers –not on the organizational ambitions and paranoias of existing institutions. The more large institutional players try to polarize the discourse around “us” vs. “them” scenarios, the harder it is to have intelligent discussions on these topics.

That being said, the more tenaciously the RIRs resist needed economic reforms in their approach to ip addressing, the wider they open the door to alternative institutions willing to promote innovative policies.

Unfortunately the materials for this meeting, including the report, are hiding behind the ITU’s TIES interface. Unless you are a member you can’t gain access to them. An academic paper based on the report, however, will be accessible soon.

12 thoughts on “The ITU looks at IPv6 addressing

  1. Milton –
    I've read the study several times, and per your summary: “This study reviews the economic rationale for IP allocation policies and argues that conservation is still a valid concern for IPv6 due to the large basic unit of IPv6 allocation.” You then go on to propose the TABL address approach as a economic solution to provide incentives for appropriate conservation. You also note that it is “outside the scope of this report to project the future of Internet routing”, despite the inherently linked nature of addressing models and routing.
    As I have noted, we don't face a major IPv6 conservation issue, but do have serious routing concerns in the future. Reference Mr. Kuerbis post-IGF IGP blog post: “What became clear in a debate with ARIN’s John Curran was that route aggregation, not conservation, is the predominant constraint going forward. ” Further, your own study acknowledges Tom Narten saying that our present IPv6 assignment model might have up to 1% of IPv6 space assigned within 50 years.
    With a trend of 1% usage over 50 years, it is actually quite difficult to argue that conservation is a prime concern, and the introduction of a new address management approach which doesn't fully consider the potential routing impacts is not a responsible course of action.
    The is why you're seeing resistance to your “innovative policies”. It's not based on some “ITU/RIR rivalry” but simply reflects the RIR systems ongoing focus on keeping today's Internet running.
    /John
    John Curran
    President and CEO
    ARIN

  2. How does ARIN know what constraints network operators face when making routing decisions?

  3. > How does ARIN know what constraints network operators face when making routing decisions?
    ARIN is composed of more than 3500 organizations which include network operators, hosting companies, enterprises, etc. These issues are discussed at twice-yearly meetings & mailing lists which are open to the public and the topic of routing implications of address policies has always been a topic of discussion.

  4. Are financial kick-backs, bribes, and other forms of graft paid to ARIN employees and Board Members for Address Space allocations?

  5. Excellent question. Our policy is that no employee shall solicit, accept, or agree to accept, anything of value under circumstances which could reasonably be expected to influence the manner in which the employee performs work or makes decisions. From a practical matter, this means that a thoughtful gift of a box of chocolates or candy will be put in the break room and consumed (rapidly 🙂 by all, but any object of material value gets returned.
    The number resource allocations are made only by ARIN staff based on documented compliance to the policies developed by the community in the Network Resource Policy Manual, and oversight of this process is provided by myself and ARIN's senior management team. ARIN also undergoes periodic operating controls audits to insure compliance with these policies.

  6. How does that apply to the infamous SWAMP ?
    Also, how does one audit situations where a U.S. Government grant (NSF?) is used to obtain large allocations
    and those address spaces are found later to be sold to commercial operators to fund community leaders (aka Union Bosses).
    Will a /16 audit be done of the IP4 Address Space to look at all 65,536 possible blocks, their history, usage, etc.?

  7. Hi, John,
    The RIRs have developed a bad habit of playing the route aggregation card whenever they are confronted with any policy option that departs from the status quo. The problem is that there is no scientific evidence, or even suggestive qualitative analysis being done that supports claims that one policy option leads to more de-aggregation than another. This work remains to be done and I'd be happy to help do it, but one must adopt a scientific attitude toward it.
    The TABL proposal, as far as logic can tell, would not have any effect on routing aggregation, and if you want people to believe otherwise, you need to provide some evidence and analysis. Under current policy, any organization in the world that wants PI space and demonstrates “need” can and will be assigned a /48. Under my proposal, any organization in the world that wants PI space and pays a recurring fee will be assigned a /48.
    You will need to explain to me, and to the world, in more detailed terms than is probably possible here, why one option creates more routing problems than another. You haven't done so yet.

  8. Milton,
    You need to pick the correct forum for this idea. it won't be decided in the the RIR policy discussion communities, the ITU or the court of public opinion. You need to write a draft RFC and take it to the appropriate IETF WG. They are the ones who tell the IANA what to do in re: IPv6 address blocks.
    BTW, when you use terms like: “Soviet-style central planning”, it just makes you look like a wanker…Seriously it's just bad form.

  9. Milton,
    Points… On one hand you propose a very open market, very Keynesian model for address space treating it as a resource (similar to shares on a stock exchange) but on the other, you propose a central organisation model where the resource is given a price by a virtual monopoly that is paid yearly for a certain amount of resource… Will you then propose the creation of state controlled agencies that manage v6 within the nation and hold auctions and then charge further amounts on a yearly basis?
    Considering the spectrum model has created serious market failures such as spectrum hording and oligopolies that then had to be and still have to be addressed and considering the model you are proposing is MORE stringent (and likely to lead to an even greater market failure in the event that v6 DOES become scarce) than that of spectrum allocation… I am surprised you even claim to talk science and economy.
    It seems to me that out of the two models, the one that exists and the one that you are proposing, the one that least resembles open markets is the one you are proposing. Under current system, those who need and use the resources are those who make the decisions on how the resource system works. Under your model, control goes to… A singular entity with sections controlled by other entities that form a part of that singular entity…
    I would like to ask you this: Do you even WHY the Internet works?
    On a completely different note while I think your Soviet comment is disrespectful, I would like to point out that your model is more like it than the RIR model.

  10. Fahad:
    I can tell you haven't read the report. The proposal specifically rejects auctions as a method for making initial address block allocations, and spends quite a few pages explaining why. And yet you are saying that I am calling for auctions.
    You say that my model constitutes a “virtual monopoly” because it charges fees for addresses — but so does the RIR system. The only difference is that for TABL blocks, there would be no “needs assessment,” just a fee. That's a lot less restrictive and monopolistic than a shifting and complex system of needs assessment. Fees are needed to encourage conservation and reclamation of addresses. My proposal simply said that the fees should go up as the size of the address block occupied increases. APNIC seems to have already done this, the other RIRs haven't.
    The TABL proposal is designed to work completely within the RIR system. It proposes to set aside a small part of the overall ipv6 address space for TABLs while the rest of the space is allocated in the traditional way. Not a big deal.
    Central planning in communist regimes was based on the idea of allocating according to need and not using the price system. I don't care whether it's respectful or not, if it's accurate and useful to make the comparison between Soviet-style central planning and the RIR approach I will make it. Maybe it will wake some people up. And you mistakenly call the proposal “Keynesian” (huh?). Try “Coasian.” That one was good for a laugh. I suppose you haven't read Keynes, either – but I don't blame you for that.
    On the other hand, asking me wherher I know “WHY the internet works” is just a childish insult, as is the previous commenters use of the term “wanker.”
    Both comments reveal the poor state of discussion within the technical community when it comes to the basic techno-economic and institutional issues of address allocation.

Comments are closed.