LIR/EU Transfer policies are a bad idea, and there are other options. To understand this we must first agree on what a “LIR/EU transfer policy” is. Then we can look into the problems with this type of transfer policy (and there are many). Such a policy slows IPv6 adoption, which is a very bad thing at this stage of the game. It also commoditizes, monetizes and privatizes IP space, space that should remain strictly a public resource. A LIR/EU transfer policy creates and encourages bad behavior in the Internet community and accelerates routing table growth. Amid all of these negatives, the most surprising fact is that there is no compelling reason in favor of these policies. The good news is that we have a number of other tools to leverage in a soft-landing strategy which can be affective.
A LIR (Local Internet Registry) / EU (End User) transfer policy allows an organization which has been allocated IPv4 address space from a RIR (Regional Internet Registry) to transfer that space to another organization which would have otherwise requested an allocation of IPv4 address space from the RIR directly. Such policies are a reaction to the fact that the IANA free pool of IPv4 address space will run out in the next 2 to 4 years, meaning that no more “new” IPv4 addresses will be available. It is generally assumed that any such policy would restrict transfers to a “region local” scope but beyond that there is a lot of room for variance. The transfer policy currently before ARIN is PP 2008-2. It was developed by the ARIN AC (Advisory Council) and contains a fair amount of complexity and restriction. I congratulate the AC on a very well thought out policy proposal, they have at the very least attempted to cover all of the problems and risks that a LIR/EU transfer policy inherently creates. My position is that all of these problems and risks that they are trying to mitigate actually outweigh any benefit that a LIR/EU transfer policy may provide. There are other transfer policies either under review or in place at other RIRs around the world as well. I will not go into more detail on the various proposals and policies here but I do encourage you to read them if you have not.
We need to move to IPv6 and the thing that is finally forcing people to get past their denial of that necessity and begin migration/activation/etc is the threat imposed by IPv4 free pool exhaustion. More people are seriously looking at v6 now than last year, more last year than the year before and so on. This is only because they are being forced to by the undeniable truth that new IPv4 space will no longer exist by 2012 (and possibly a lot sooner). A transfer policy gives them a “way out.” It allows these people to sink back into denial by telling themselves: “Yeah, the IANA free pool will go away but I can always just go buy/rent some addresses – no big deal, no need to worry about it.” Everyone will not take this stance but certainly a great enough number to cause an overall slowing of IPv6 implementation will. An IPv4 LIR/EU transfer policy is a public statement by the RIR in support of continued IPv4 reliance. It says that IPv4 is something that we can not live without, that we need to take drastic action to ensure its availability. This in turn is an implicit statement against IPv6 development and adoption at a time when the exact opposite message needs to be made clearly and loudly. Slowing IPv6 adoption in any way hurts the entire Internet community. Once the majority of organizations begin v6 adoption a catalyzing effect will help to ease everyones progress. When more people demand IPv6 products, more will become available and at a lower cost. When more people demand v6 content, more will be produced. When more v6 content is produced or made accessible, more people will demand v6 access; which will lower access costs and make it more available. All of this will create more skilled IPv6 developers, engineers, etc. Once v6 content, access, knowledge and equipment is available at a reasonable cost, innovation will explode! That is when we will see the benefits of IPv6 truly realized. This will happen eventually, there is no choice. The question we face today is when will it happen and how much pain will we all incur on the way. Delaying IPv6 will increase the pain, of that there is no doubt.
A LIR/EU transfer policy creates a legitimate market where a dollar value is assigned to an IP address in a tangible way. This is something that has never been the case before and many would argue (including myself), should never be the case. IP space (v4, v6, vX) is a public resource and as such should be borrowed, used and returned by those with a need for it. Think of it as a library book; when you need that resource you are free to borrow it but only with the expectation that you will return it when your need has elapsed. Attempting to sell or rent it to another is an abuse of the system and would never be allowed. Now assume that the planet runs out of paper from which to print books, does this change anything? I say NO. You still borrow the resource when you have a need for it and then return it for others to use when you can. And what if all the books are checked out? Lucky for us there are e-documents now available for you to borrow and as we see it today, they can never run out. You have an option, just as we have IPv6 as an option and just like the library that never runs out of books because people accept e-documents; once IPv6 is accepted, there will not be a shortage of IPv4 space.
There is also the question of how IP space trading will work, how a buyer and a seller find each other. I see two options. One is that ARIN provides a listing service, the other is that private brokers emerge. If ARIN creates an IPv4 listing service, they maintain a greater degree of control but further ingrain themselves in the assigning of private value to what was (and should remain) a public resource. This puts them in direct cooperation with the molestation of the current system and strengthens the statement discussed above. It also creates at least some more work for the staff but only temporarily. This is a problem because you can’t say how long it will last so you end up hiring folks just to fire them once IPv6 adoption becomes the standard and v4 space loses value. If ARIN stays out of the listing business, they lose a lot of control. In this scenario, you may be able to stop speculators with a well written policy but I do not know that you can stop profiteering in the broker business. There are countless opportunities for price fixing and other corruption if ARIN does not regulate the trade.
By implementing a LIR/EU transfer policy, the RIR is inherently raising the value of IPv4 space while also causing the IPv6 infrastructure build-out to slow. This combination could have a detrimental affect on small businesses who (potentially) will not be able to afford the now costly IPv4 space and will thus only have access to an underdeveloped IPv6 Internet. Small businesses are often where much of our innovation and growth happen so this monetization of IPv4 space could have a seriously negative affect on the economies (at least in the CIT sector) in that RIRs region. While I am not an economist, I think that this is a strong enough possibility that it should be considered when weighing the issues.
Announcing an IPv4 LIR/EU transfer policy provides a direct incentive to organizations to hold onto their unneeded and unused IPv4 space rather than returning it (as they should). When faced with the prospect of legitimate profit, most organizations will wait for the market to emerge and once it is in place will wait for the price to go up. This IP space hoarding ensures that the RIR will not recover as much space as they would have otherwise. Most in the Internet community believe that as we near free pool exhaustion, there will be a tipping point which triggers a “bank rush” or “land grab” situation; a period where organizations rush to take the remaining space. A LIR/EU transfer policy has the very real potential of making the “bank rush” happen quicker and more forcefully. People will want the space before they have to pay for it and more people will want space simply to hold onto and sell once the price goes up. All of this will actually speed free pool depletion and is directly counter to the goals that such a policy is attempting to achieve, a so called “soft-landing.”
Routing table growth is a very real problem already and explosive routing table de-aggregation due to IPv4 transfers will make it unbearable. Without some form of market manipulation there is no real way to prevent or restrain it in a transfer paradigm. Even with intervention, there is little reason to believe that it won’t happen. This is because the very nature of a LIR/EU transfer policy encourages organizations to de-aggregate. It encourages IPv4 holders to break off a network from their aggregate to sell/rent to the highest bidder. Brokers who make money on each deal could profit more from a greater number of deals by brokering smaller blocks. Other organizations who have space available will quickly realize that they can keep prices up by releasing space a network at a time, preserving scarcity but driving the routing table count up.
The IPv4 black market that many are worried about forming after free pool depletion is actually already here. I have seen it, I know that it is happening. I can not know to what extent but I have personally witnessed deals being struck for the use of IPv4 space as have others. The fact that it exists today when v4 IP space is free and available, illustrates that a transfer policy will not stop it in the future. People who are willing to break the rules do not care what rules you put in place. A LIR/EU transfer policy actually has a greater chance of making this black market larger and stronger than it does of curbing it. This is because the public trade will help establish value and cause confusion. If IP transfers are strictly prohibited, it is harder to mask a transfer and harder to set a value. Furthermore, any black market that does exist will mostly solve itself as IPv6 adoption occurs; the more of the Internet that becomes IPv6, the less v4 space will be worth. Therefor, the most effective way to stop the black market is to speed IPv6 proliferation, not allow transfers.
What is needed is a comprehensive program of soft-landing policies. Most in the community understand this, the evidence being that much of what should be done is already underway in some form. The first thing I would recommend is to abandon all IPv4 LIR/EU transfer policies and make a public statement that there will be NO transfer policy, ever. This will help those with underutilized space to return it voluntarily by removing the gold at the end of the rainbow. Next, ease and incent the return of unused legacy space as much as possible. There is a current policy proposal with this aim, PP 2007-17:Legacy Outreach and Partial Reclamation. Incentives for the return of all unused space should be made as strong as possible while keeping the routing table in mind. Coupled with these incentives (the honey) RIRs should also implement more aggressive auditing practices and stronger punishment for not returning space, using space improperly, etc (the stick). Imposing fines could help cover the cost of audits, etc. There is a policy proposal focused on the auditing aspect; PP 2007-14:Resource Review Process, but its wording needs to be cleaned up some (yes, I am volunteering if the need is there). Going back to the library book analogy, a policy which sets a time frame on IPv4 allocations may make sense now. When you check out a book, you are expected to return it on a certain date. If that date comes and you are still actively using the book, you are allowed to state that and keep the book. Since we are at a point now where IPv4 space will start to be replaced by IPv6 space, it may make sense to implement a similar policy at the RIR with regard to IPv4 space – that is a time frame. This policy would require that after X amount of time, the LIR/EU would need to return to the RIR with justification if they wish to keep the space. The burden should be on the LIR/EU to prove that they are actively using the space. Again, I am volunteering to write such a policy if others feel that it is a good idea. One last thing that I support which is already underway, is to create an RIR-to-RIR transfer policy (ARIN PP 2007-27). [EDIT: this policy proposal was abandoned] This limits some potential corruption (shell companies, etc) and lets the “busiest” regions IPv4 pools last a bit longer without the statement of support for IPv4 that a LIR/EU transfer policy produces. It also spreads the benefit of legacy IPv4 space reclamation across the globe going forward.
An IPv4 soft-landing strategy is vital and there are many policies which could be successful if properly combined. A LIR/EU transfer policy is not one of them. The focus at this point must be much more on promoting IPv6 than on selling IPv4.