CircleID

Brilliantly written article, there.

But, on the topic of the commons, and overgrazing of them, the alternative was much worse - pastures enclosed by landlords, and let out on a share or rent basis to tenants, who were then at the mercy of the landlords.

With a benign and enlightened landlord, this arrangement worked very well indeed. With absentee, rackrenting landlords who stayed in London or on the continent [well, Europe], playing cards and drinking in fashionable clubs, just sending an agent once a quarter to extort as much rent as he could, the arrangement was a spectacular failure.

Much the same with the conflict between homesteaders / settlers and ranchers in the old west, and the role that barbed wire fencing played.  There, there was a conflict between possession of huge amounts of land to feed a comparatively small number of free range cattle, and a crowd of homesteaders whose cropping techniques swept away the thin layer of grass holding the soil together through most of the midwest, producing the famous dust bowl, the resulting drought driving several families out of their houses.

IP address allocation and management, compared to land ownership and management is an interesting study I'd say. The parallels become immediately obvious, and we are rewarded with a lot of history to study, that, with careful policy making, need not necessarily repeat itself.

I don't think that the word "competition" is really the right one to use; in the context of IP address allocation it is not a word that I would use.

I agree that rational allocation of IP addresses is important.

However, your note glides over the extremely difficult issue of deciding whether a claimed-need for address space is "genuine" (your word) or not.  What is a "genuine" need to one may be considered frivioulous by another.

That decision is today made using policies that give more weight to quantitative technical metrics than they do to softer social and economic metrics.  And there is not a clear articulation in those policies of the desired balance between short-term goals and long-term goals.

We don't really know the total effect of the RIR and ICANN/IANA IP addresss policies.  My own sense is that at least some of the drive for deployment of NATs comes from the perception that it is easier to deploy a NAT (and thus break the end-to-end principle of the net) than it is to obtain address space.  In other words, it may well be that the attempt to conserve and aggregate that is in the existing IP address policies is causing a subtle erosion in the end-to-end quality of the net.

In any event, these are difficult choices - and I believe that the RIRs have been slowly working through to answers that are adequate if not perfect.  My primary concern about IP address allocation is that it is perceived as sufficiently arcane that those who are more distantly affected by these policies tend to refrain from engaging in the making of these policies and, as a consequence, the softer social and economic aspects of IP address policy tend to be under explored.  And at the ICANN level, IP address issues seem to be an institutional blind spot.

As for "competition", or rather as for the idea of creating per-country pools of IP addresses.  From a technical perspective that seems like a bad idea to me - it has usually been my experience that in most cases dividing a single pool into multiple pools leads to less efficient use.  My own rule of thumb is that a pool of resources should not be split unless there is a specific concrete justification for doing so.

To make a difficult issue even more difficult, in the case of IPv4, I do not believe that we have sufficient total space to absorb the impact of the inefficiencies that would be caused by per-country pools.

But with IPv6 with its much larger number of addresses the answer might be different.

While we are discussing the fragmentation caused by per-country pools, it is often overlooked that the RIRs form per-continent pools and that the same kind of fragmentation and inefficiencies are caused by having multiple RIRs as would be caused by having per-country IP address registries.  The difference is merely of the amount of inefficiency (less) of having the existing continental RIRs versus the amount of inefficiency (more) were there to be per-country IP address registries.

So in the sense that per-country registries are simply a more fine-grained system than we have in the existing continental RIRS, the difference may not be a difference in kind but rather simply the difference between two points on a continuum.

--karl--

For some more perception on competition and adam smith's economic theories, try Tom Vest's paper "The Wealth of Networks", http://www.pch.net/resources/papers/the-wealth-of-networks/Vest-WoN@SIMS-DLS-050316.pdf

> quantitative technical metrics than they
> do to softer social and economic metrics.

Nothing much more than "I've made the best possible use of the IP space I have, I plan to use so much more IP space in the coming year, so please allot me some IPs". 

People deploying NAT and other people pointing to this as evidence of the shortage of v4 addresses - NAT deployers dont typically get IPs direct from the RIR.  They're a much smaller operation who gets IPs from their ISP.  If the ISP is stingy or excessively bureaucratic about assigning IP addresses

This situation is quite likely to be the case, especially when you consider the customers of several incumbent telcos in china / other parts of asia / africa - places from where most of the complaints of the we're so short of IP addresses we HAVE to use NAT" originate

Per country IP registries - what does exist is LIRs - local internet registries like JPNIC for example, or CNNIC, that receive IPs from APNIC and take on the task of distributing IPs in their country, but according to the APNIC framework (including requirements about justifying IP addresses, etc)

In the ITU v/s RIRs case being discussed, it is more of a turf war / battle for control, where countries want to just be allotted IP space, which they can manage and distribute as they see fit, with or without regard for the frameworks already in place.

Geoff and Suresh have said it well - this is essentially a blind power grab by the ITU.

You can't have competition amongst what are basically regulatory organizations - you just get more regulators. The RIRs are membership organizations, not cartels. The members can vote to change prices and policies at will.

A more radical change, such as allowing the trading (buying and selling) of IP Address space as a commodity would have much greater impact on the availability of address space. If the ITU is truly in love with the invisible hand, let them embrace it by convincing the RIRs that address space should be a fungible commodity.

Geoff, Paul,
Well written article.

I have heard the following arguments at various times.  Doubtless, so have you:

* IP addresses are hoarded by "developed nations" - if only "underdeveloped" nations [replace with similar word] were given more IP addresses, the Internet would grow more/better

* Since a regional IP allocation strategy already works, there is no technical or principle-based reason to not have it work on a country basis

* IP addresses are a national resource; no small group of geeks should control it

* With the huge number of IPv6 addresses available, there is no mathematical or logical reason to create an "artificial scarcity"

* The ITU helped create a smooth, well done national telephone system worldwide, with clean bridges nation to nation.  And this was done decades ago

* This nation based phone system has brought billions of dollars of economic prosperity to the nations, allowing nations decide the rates and tariffs themselves and avoiding undue regulation.  Address registries are keeping nations from their legitimate economic due

* Each nation will have power and be able to govern "their" IP space in their own national self-interest, without interference from other global bodies

I don't believe there is a clearly articulated, well-written response that responds to many of these statements.  Without an appropriate response, I fear that many administrators of Internet policy in many nations will find far more resonance with the appeal to the nationalistic instinct than the appeal to rationale and global addressing sanity.

Ram, here are a few answers, sort of -

A lot of the points you raise are covered in several papers available on NRO / the apnic website's public policy section etc, but ...

1. It is a case of demand and supply, and any entity, anywhere, can get IP space as long as they can justify their current utilization.  Best of all, I'd suggest showing Geoff Huston's BGP movie to skeptics

2. Economies like China are currently some of the largest recipients of IPv4 IP blocks, and large netblocks have been allotted to AFRINIC and LACNIC for distribution, in recent times.

3. The telephone number system, for all its complexity, is nowhere near the size, magnitude or costs of the internet.  If you were in india in the mid 90s when the 'net was first introduced, you too will remember paying INR 15K for a flaky dialup, and very high local call costs increased by the number of times you had to redial because your modem dropped without connecting.  If the internet is supposed to work the same inefficient way the phone system works, and is subject to the usual extortionate settlement charges that form the bulk of developing country telco earnings, and absorb a whole lot of their losses ... [now guess who the strongest proponents of settlement based peering are?]

-srs

Suresh, I agree with your perspective; however, in some nations, this perspective is far less attractive than the nationalistic one - witness the public statements made by various government folks at the last WGIG meeting in support of nation-based IP allocation policies.

In addition, conventional wisdom is that the phone network just "works", not that it is inefficient; similarly, the Internet is often defined as "work in progress" (implying not as stable).  More eduction is needed to chip away at this conventional wisdom.

I read the session transcripts of that WGIG meeting that have been posted on www.wgig.org, and am gratified to know that Leichtenstein (on behalf of the presidency of the EU), Australia et al do understand the issues involved.

Most of the people making those comments (especially the representatives from India and China, both of whom I have had the pleasure of meeting at various other conferences) are, with all due respect, not aware of the true technical or even social / economic complexity of the issues being discussed, and nor are they really aware of the true nature and governance of the RIRs.  Oh, I am glad to see that the delegate from Syria was in full form. He is, how shall I put it, quite well known for his frequent contributions at other ITU meetings :)

Here is my perspective on a few issues raised there ..

APNIC, for example, is governed by its AC / EC, and members from any APNIC region can, and regularly do, stand for election to these committees.  This, and other RIRs, are a cooperative undertaking by ISPs from around the region of coverage of the RIRs, asiapac in the case of APNIC. 

ICANN, both the board and its At Large component, is fully open to people from any country, anywhere, standing for election and making their opinions known as well, besides helping them gaining some better understanding of the underlying processes.

So, raising the "other" invisible hand - the invisible hand of US Government remote control - is not going to be productive or useful.

Nor are the usual claims that "China is short of IP addresses" - China has been receiving ever increasingly larger blocks of IP space in the recent past, and all IPs there are allocated by the LIR, CNNIC - which shares the apnic common pool of addresses, but is responsible for allocating it locally.. so any shortfall of IP space that China raises in future should result in them being referred right back to CNNIC, which is a quasi governmental agency, I believe.

Then there is the "holders of international bandwidth are at an advantage" theme that gets harped on - this time by India.  Some people in India, and others active in the asiapac network operators conferences, have been advocating the need for an internet exchange long before NIXI was formed.  However, the way NIXI took shape and is being run is far below potential, and improving this (plus adding an Akamai cluster, and mirrors of popular download sites, colocated at the datacenter of any of the ISPs that peer at nixi) will go a long way towards improving this situation.  Oh, I almost forgot. India's input to the WGIG said something about root servers all being located in developed countries.  I am sure now that they've retained Afilias for the .in ccTLD they'll have some slightly better knowledge of anycast.  Put it to them that the root nameservers are anycast, and mirrors of (say) I Root or F Root would be an excellent thing to have at the NIXI [or at an isp that fully peers, advertises all its routes, at nixi]

I could say more, but these issues have been pointed out in the past, and right now, I am concerned that the effort to promote internet governance is being superseded by an effort to wrest control of the existing structures from their incumbent operators, who hold these in trust for the international internet community and operate it as directed by the community.

This is not a war of independence against a colonial power, though it appears to be projected as such.

If governments wish to participate in the governance of the internet, the existing framework is extremely accomodating and can fit them in quite easily.  Government owned telecom and internet providers (BSNL / MTNL in India, China Telecom in China etc) can, and should, play a larger role in this, by standing for election to the APNIC AC for example, and actively involving themselves and contributing their expertise in other ways at APNIC.

-suresh

Tom Vest just posted an excellent circleid article on this subject - http://www.circleid.com/article/1064_0_1_0_C/

Required reading, at least by the indian and chinese representatives at WGIG, before they continue to advocate the Zhao proposals.

In response to Suresh Ramasubramanian:

You wrote:

"ICANN, both the board and its At Large component, is fully open to people from any country, anywhere, standing for election"

My opinion and experience (I was the only board member ever to be elected to ICANN to represent North America) are quite to the contrary.

ICANN has eliminated all means through which people from any country can stand for election for any role in ICANN that is more than that of a mere observer.

ICANN's "ALAC" to which you allude requires that a person join an ICANN approved club.  That club must, in turn, join a larger ICANN approved regional club.  That regional club can name a few people to yet another ICANN approved club. And that regional club can name a few people to a global, but still ICANN run, club.  That global club can then name a few people to a nominating committee.  That nominating committee is then diluted by a strong component of industry designated members.  Eventually this last club gets to appoint a small subset of ICANN's board of directors, the rest of the directors coming from pre-selected industry segments. 

The system that ICANN has established for the public to participate strangely resembles the old system of soviets (committees) that existed in the old USSR.  And the net effect - that of insulating the decision making forums form the public - is about the same.

On another point, you mention the lack of technical expertise on the part of some WGIG participants.

It is easy to dismiss the people who are working on the WGIG issues because they lack technical credentials.  But one must realize that most people from the technical community are equally ignorant of matters of law, economics, and governmental processes.  For those reasons, it is easy for those on the non-technical side to dismiss the opinions of those with more technical backgrounds.

Karl

I wont comment on ICANN here. But in response to this -

On another point, you mention the lack of technical expertise on the part of some WGIG participants.

> It is easy to dismiss the people who are working on the WGIG issues
> because they lack technical credentials. But one must realize

I rather agree, that is a valid point.  In response to this I'll just repeat what I said at an igovernance panel organized by apdip/undp during apricot kyoto ..

http://www.apnic.net/meetings/19/docs/transcripts/igov.txt

The ideas are mine. The words are mine. The english as transcribed in that link sure doesnt sound like mine though, I guess I can put it down to transcription error :) Anyway, here goes.

I have filled in one or two blanks left in transcription, from my memory of what I said there. And corrected some of the english as well.

---------

So here what we have is technologists appointing to poor misguided regulation saying that isn't going to work or pointing out that the Internet has no national boundaries so traditional regulation is not going to work.  On the other hand we have governments insisting we do need a route because we have substantial interests that are being affected by this.  For whatever motive. To preserve a national monopoly that is bringing substantial money to the government for example an income in telco.  Or for other even more mercenary reasons such as a plain desire for control - or from a sincere desire to do good and ensure good for the people by good governance.  So where and how do we channel
the energies of governments towards good governance where they can do the best.

For example, you have a lot of issues that depend on international cooperation between governments, and between governments and industry.  Cyber crime, for example, and anti-spam regulations - wherever there is a crossover between the online world and the offline world, and where participation and cooperation between law enforcement / judiciary / governments / ISPs around the world is essential in order to get the job done.

In the course of my job I work with a whole lot of governments, besides working together with other ISPs and businesses.  And I keep wishing there was already a framework in place instead of trying to build it from scratch at ITU/OECD meetings, as well as at network operator meetings like the ones I help organize at apcauce.org

In the case of the telecom sector and inequitable pricing of telephony and bandwidth in developing economies, governments could help localise the industry [by helping set up internet exchanges], arrange / broker appropriately equitable prices with other providers. 

That's what I was trying to say.  How do we get governments to participate in the current governance process, without undermining the foundations of the existing process, and without unintentionally making their participation intrusive or harmful?