These are unedited transcripts and may contain errors.

Address Policy Working Group
2nd session
27 September 2012
11:00 a.m.:

GERT DOERING: So, enough coffee? Welcome back everybody. And welcome all those that have been to IPv6 in the first slot. I don't see Jan yet. He asked for to have a discussion here and he is not here yet. But he will show up, I am sure of that.

Anyway, this is the ?? We Have Addresses Left Policy Working Group, and this is our agenda for the second time slot today, which is basically discussion of these three open policy proposals. There is some challenges to it because the proposer of 2012?06 and 2012?05 both can not be here but have volunteered to call in by Skype with video and everything and when we tested this yesterday it worked out very nicely. Let's see whether it works out today.

The first thing is 2012?06. It's one of those twists that remove bits of the policy that have been necessary due to IPv4 runout and now are just getting in the way. Tore Anderson was the first one to notice and actually do something about it and he is here on Skype.

CHAIR: You are on screen now and if you share the presentation, we can see you. The floor is yours. And does this work? Like so?

TORE ANDERSON: Yes. And okay. So, my name is Tore Anderson, and I would like to be there with you in Amsterdam, but unfortunately I am sitting here in my basement in Oslo, so welcome to my basement.

Anyway, the policy proposal is very simple. It's to revert runout fairly, which is another policy proposal, after IPv4 depletion, and as you all know IPv4 depletion has happened already, so the proposal right now is simply to revert runout fairly.

That is all it does. It reverts the exact text that was added by runout fairly, and puts back whatever was removed. It does not add any ?? not a single phrase, word or anything that was not in the policy before, so it's a direct unpatching of the policy you could say.

So, the reason why I think this is easy enough to justify is that runout fairly, it did its job, we already ran out, so whether or not it did its job in a good way, if it was a success, I want to try to be the judge of that, but it has no purpose any more because we have run out and it's the allocations from the last /8 is what, is providing any fairness at this point in time. And the NCC does not evaluate need any more, it's a /22 or nothing. That's it.

So, why do I bother? My primary reason for it is that the runout fairly proposal also lowered eventually the assignments window for PI assignment either to the LIRs customers or to the LIRs own infrastructure to three months. And I was seeing Alex's presentation and it said it was three months, and that was the half?way through time, and I just want to clarify that half?way through the time that you have to use half of your assignment is the half of those three months. So, six weeks. And I know because when I sign a new customer, and I order hardware and stuff like that, it might even take six weeks before the hardware has arrived, much less than I have been able to unbox it and install stuff on it, which means that I can't actually assign anything to my customers under this policy.

So, Nina also pointed it out, it's much too low for normal business operations.

Another thing is that runout fairly lowered need period for transfers of allocation to say three months. And I don't think that was quite intentional. It was just something that happened because the transfer policy doesn't really specify its own time period. It just relies on the NCC to LIR period. And now that we have run out, I am guess that go people are more interested in transferring space and three months again is not nearly enough time for normal business planning.

And so these two effects are still active, but they have no impact on the use of the allocated pool or fairness because the pool is empty anyway. So, in my opinion, these remaining residual effects are negative and we should just get rid of runout fairly in order to go back to how it used to be before.

So, and also, there is the benefit that the proposal just removes some out of date text. It is a reduction of policy text in total and especially those slides that be thereafter this date something after another date, something else, and then the reader of the policy has to kind of read through all of these and end up with what is actually a valid policy, so it improves the readability slightly. As I am sure Rob is going to be talking about the IPv4 policy document is still in need of major amputation to actually get back into a state where it only describes relevant stuff.

A final point I would like to bring up is that there is also another policy that Sander presented that wanted to increase the transfer ?? I have lots of feedback ?? this is difficult ?? I am going to take off the headset.

So, Sander's proposal seeks to increase the transfer period to 24 months. And that does not conflict with my proposal to revert runout fairly because as I mentioned earlier, the current transfer policy does not specify a time window for need for transfers. It just refers to whatever the NCC to LIR time period is which was lowered to three months. And Sander's proposal will change or add text straight into the transfer policy section. And if that passes, my proposal will not actually change any affected policy, it will be a no op basically, except that it will remove some, or reduce the amount of out of date and not relevant text in the policy.

So, that's my proposal. There was not that much feedback on the mailing list, but mostly positive. I don't think that ?? I didn't interpret it as anybody actually was objecting to the proposal at all, and there was mostly positive feedback, but of course, I am expecting that at least Nina from TDC should also state support for this proposal because it's exactly what she wanted in the previous segment of this Working Group.

That was all. So... thank you. Any questions?

CHAIR: I am moving over here so Tore can actually see me there. There is a web cam there so Tore can see you. Thanks for bringing this up, and volunteering to be on Skype. Now, turning to the room, so, I see Remco.

AUDIENCE SPEAKER: Remco. Good morning, Tore, thank you very much, I think this just makes perfect sense and I strongly support your proposal. And thank you.

CHAIR: Thank you. Other comments on that? I take the remark from Nina this morning to support this policy proposal because it's basically solving the issues that you brought up. Okay. I don't see anybody actually speaking up against the proposal, and as we had support on the list and no objection there, I think it's in the discussion phase and at the end of the discussion phase we can go forward with it, so ?? and and fantastic.

CHAIR: That was an easy one. I think we might have to go to lunch early today, but let's see how the next discussion goes. Thanks Tore again.


So, now we are back to the PI policy proposal that was on the agenda for the first slot this morning and was requested to be brought here. So, don't worry about the sequence of letters in the alphabet. They are traditionally weird and today it's even more exciting. Nick Hilliard has brought this up. There has been interesting discussions on the mailing list and I think there will be discussions today here.

NICK HILLIARD: Hello everybody. My name is Nick Hilliard and I am CTO of INEX in Dublin, and I want to talk to you about PI assignments from the last /8.

But the first thing I need to talk to you about is actually an element of jargon because when we talk about the last /8, in fact we're not talking about the last /8. What we're in fact talking about are all of the leftovers of the RIPE address space, and this is a couple of components. It includes on the one hand, the last /8, which is, and on the other hand, it includes all of the other reclaimed space which is going to be coming into the RIPE NCC, you know, as time progresses. And there is a small but a relatively constant amount of IP address space that's coming in, it happens when you have an LIR which closes or for whatever reason there might be a bankruptcy or if it's assigned to a person and that person dies and there is no clear succession policy, whatever, then that address space will naturally revert undercurrent policies back to the RIPE NCC for reallocation. So the last /8 is a very misleading term, because it implies that there is just a fixed amount of IP address space and that we're dealing with a short?term problem. In fact, we're not. We're dealing with a much longer term problem which will extend as long as we actually have IP address space in the RIPE service region.

So, the question is, who gets the slice of the pie? And at the moment, we have the last /8 policy, which is the remaining address space policy, and it deals with three things. It deals with allocations to LIRs. It deals with assignments to IXPs and there is also a reserve area of /16 for unforeseen circumstances. We don't know what they are. It would seem like a very good idea to keep a little bit of address space on the side. But, you know, hey, it's probably going to be useful at some stage. The problem is that the policy does not deem with PI assignments and because there is no policy to deal with PI assignments, that means that PI assignments from this remaining IP address space are simply not possible. And this is quite serious for end users because if you are an LIR or an IXP, you can get a small slice of the pie, and if you are everybody else, you get nothing. Which is pretty harsh.

Now, the RIPE NCC has performed an Impact Analysis on policy 2012?04, and they noted a bunch of things in the Impact Analysis. The first thing was that it was going to cause a much faster runout than we might otherwise have. It was going to create the facility for getting cheaper IP address space, because you know, effectively €50 for a /24 is going to be a lot cheaper than opening up an LIR or you know, getting existing address space or getting new address space from your existing LIR. And they introduced the notion that there was going to be an expectation of abuse in the process of acquiring IP address space or at least acquiring PI address space and this is certainly a problem.

But, on the other hand, we have no PI policy and because of that, this is a problem because we may attract regulatory interest. Because all of a sudden, we are now in a situation where we, as most of us are represent LIRs in this room, and we're creating this policy which gives us all of the rights to all of the address space, but it gives no rights to all of the other 700 million people in the RIPE service region to say that they also have the right to get some sort of portable address space from the remaining pool.

And this is actually fundamentally unfair, because it damages the interests of end users.

I think that there is a serious problem here. Somehow or another, we need to come up with some sort of mechanism fora signing a slice of the pie so that it benefits not just the LIRs, but also the end users. There is been quite a bit of discussion, quite a bit of heated discussion on the mailing list. The existing policy that's on the table, some people agree with it. Probably about 50% of the people that have commented on the policy disagree with it. So, it's clear that we don't have consensus on the existing policy but if we don't get some sort of consensus on some sort of modified policy, we are going to remain with the current situation where end users do not have any ability to get portable IP address space from the remaining pool, which is not a good situation to be in. I think this situation needs to be resolved one way or another and I would invite at this stage your comments on the matter. If anybody has any questions or comments or positions that they'd like to talk about, I am all ears.

AUDIENCE SPEAKER: Wilifred Woeber, and speaking as an LIR manager. I tend to agree with the goal you want to achieve because I agree that the situation that we are having right now is pretty much unfair. There is also, I think, a good reason to think about supporting it because at the moment we have the possibility to become an LIR, which involves just putting money on a table on one hand and it gets you a /22 irrespective of the amount of addresses you are going to need. For the PI stuff, we had sort of the minimum size to be given away /24, so I do see a potential that someone who really needs it, or who really wants it, just signs up as an LIR and gets potentially three times as much address space as the project would require, which is against what we want to achieve sort of to have a slow trickle and to serve as many people as possible. That's sort of in favour.

The second comment that I wanted to make is that in your slide of the components of the address space, we are going to manage for the future, there is one component missing and that's the addresses coming in again from the IANA by way of this allocation of the reclaimed pool of addresses. Just for completeness, I don't think it changes the picture.

The reason why I'm not supporting that at the moment is not because of the goal. What we want to achieve, but because of the problem that I am missing quite a few additions to the policy, like, as it is worded right now, and please correct me if I read it wrong, could I apply for PI and I could give it away by transfer almost immediately? Because there is nothing which prevents it in the other set of existing policies.

NICK HILLIARD: Okay. A couple of issues there. First of all mthe issue of the IANA address space. You are correct, I didn't actually mention it, that was in the mission, but it doesn't substantially change the policy.

Regarding the transfer issue. I don't think there is any policy which allows transfer of PI assignments at the moment. So, this isn't an issue at the moment. However, I think it's likely that in the future, that there will ab policy to allow transfer of directly assigned IP address. So, there is possibly going to be a risk of creating a situation where people go in, you know, register companies or whatever it is, get a chunk of address space, maybe hoard it and sell it on at a later stage. There is absolutely a risk of that. I genuinely don't know if there is any way to stop this. You could certainly restrict the ability for companies to do this by pricing ?? by using pricing as a mechanism, but using sort of legal terms as a mechanism of saying well look, you know, one organisation is entitled to one /24, but that's not going to work because you can register a limited company in many jurisdictions for, I don't know, €50 or £50 or whatever your local currency units are.

So, you are absolutely right. On the one hand there is a requirement for end users who have a legitimate reason for requesting address space to be given that. But on the other hand, we do have some level of obligation to ensure that this system isn't abused to the greatest extent that we can prevent this abuse.

WILFRIED WOEBER: Just the last sentence, I'd like the community to sort of think about the overall picture like they are probably going to see transfer possibilities for legacy address space, we are probably going to see transfer mechanisms for PI space, in general the old one. If we are going to add just another special case, I just want to have this sort of awareness that the whole set of things that we are building should, in the end, form some sort of reasonably consistent overall system. Thank you.

NICK HILLIARD: Exactly. The issue there is what colour is your IP address space? Well it doesn't really matter what colour it is, it's just address space.

AUDIENCE SPEAKER: Nick, although I really understand where you come from and can to some degree, agree on the fact that it may be not fair however, the current policy, and the reasoning behind it for the final /8 was basically we're out, and there is a reservation for new companies, new LIRs to actually have address space available that they might meet once they set up their own infrastructure. By allowing this, we will basically stop those companies in the future, in the near future, to be able to start up their company and actually gain the IP address that is they need to actually get start with their businesses for their infrastructure. And for that particular reason, and the fear that I have and with me is others, that by allowing PI space in the final /8, even though it might be some reclaimed IP address which I think is going to be a minimal amount of addresses, it's going to be a major pain for the NCC in the registration service to say actually do this fairly and actually basically come into a phase where they need to see, you know, how are we going to do this? Basically running a full operations similar to what we have done in the last three weeks, and I think that it's not in the best interests to actually go forward. So, in short, I will not support this.

NICK HILLIARD: Okay. Thanks for your comments Eric. There are a couple of points there. The first point that you make is that our requirements as LIRs are more important than an end users requirements as an end user, and I think there would be a great many end users in the Internet community and the RIPE community who would strongly disagree with that simply because they are not LIRs and maybe they don't have the interest or the requirement to become an LIR. Maybe they don't need the IP allocation.

With respect to whether this is fair, there is no such thing as fair in an era of address starvation, but there are degrees of unfairness, and I think probably the situation that you are proposing, which to say that the LIRs get all of the pie and that the end users effectively get none of the pie, it doesn't really work for me.

AUDIENCE SPEAKER: Like I said, by making the change in the policy as it is, it doesn't change the game. Even an end user can register for an LIR. Whether it's debatable if they need a /22, which is the current policy, that's something that I would say, you know, that's something that we can discuss, so that it will actually be, you know F they only need a /24, that they would get a /24, but to get cheap address space as cheap PI, because of fairness, I don't think that is a business requirement. If an end user really needs IP addresses they can become a LIR because basically you know the game is, we're out.

NICK HILLIARD: I mean, essentially, there is a balance to be struck here. On the one hand, if you're an LIR and you somehow as a policy ?? I mean, I would foresee as a long term proposition that we're going to get into the business of splitting up all of our allocations into as slow /24 eventually. So if you are an end user, you register yourself as an LIR, you get /22, you sell off three /24s and you keep one of them for yourself, which means that you get the IP address space for a relatively low amount, possibly even free. I don't know. I don't ?? I don't think that the, what you are proposing is balanced. I think that certainly the proposal that's on the table is not going to work. The RIPE NCC have indicated very clearly that there are too many avenues for abusing it. So I mean, what I want to see is maybe instead of sort of saying well look let's chuck out the idea of PI entirely and hang the end users, what I'd like to see is some sort of mechanism of coming to some sort of arrangement whereby through some mechanism we can sort of say, well, look, if you need IP address space, portal address space, you don't have to get as much as as a /22. You can you can get your /24 but that we do this in such a way that it's not going to discriminate unfairly against end users or assign them too much IP address space which are they are not going to use or which they are going to flog on the grey market whatever. I think there is a middle ground between our two positions and that's what I want to work towards.

AUDIENCE SPEAKER: I don't disagree that it's, how will I say this ?? that it's an ideal solution. However, I think that the risk of running out on a much faster rate by allowing PI is larger than the actual abuse of actually setting up a new LIR. The system, and the policies that we have in place for setting up new LIRs is happening, or will happen in the near future anyway, and by allowing PI in that final /8, we will only, you know, ramp up in the use of the final portion of IP space that's reserved for the new LIRs. So, you know, neither of the two options are perfect. However, I think that the thing that you are proposing here is far worse than the abuse that we're likely going to see from the setting up the new LIRs and transferring. Is my understanding.

NICK HILLIARD: Ultimately, the more organisations to which address space is available, the faster runout is going to happen. Runout is going to happen. We can't avoid, it it is a fact.

AUDIENCE SPEAKER: It's already here.

NICK HILLIARD: We have a reserve, and we are running on reserve and it's going to run out some day. We are going to be trundling along the bottom with local returns and IANA returns. We can't stop that. We can change that. I'm not sure if changing the speed has huge amounts of merit one way or another. Anyway, let's take this discussion off line because there are a couple of people who have other comments.

AUDIENCE SPEAKER: Carl from RIPE NCC. I have a question from James Blessing on Jabber. Would imposing a time limit before transfer not be the simplest solution to that? I think it's in response to Wilfried's discussion for transfer?

NICK HILLIARD: It changes the economics because if you get your IP address space, you are going to be sitting on it for a little bit longer so you are talking about a higher rate of depreciation. I don't think it changes the policy substantially and I don't think it's necessarily going to fix any of the problems that we're dealing with in any meaningful way. Having said that, I haven't actually sat down and thought about it, so ?? but maybe, if James could take this to the mailing list and put it in, we can throw this idea around a little bit.

REMCO: So, before my own comments, a quick response to James, that's actually engineering for a problem that doesn't exist today because the only address space you are allowed to transfer in the current policy is PA. So, talking about transferring PA /PI is ?? I don't think what we can have the discussion some other time.

So, first of all, I would very much like to applaud you for daring to stand up here and touch this subject. I wouldn't have touched it myself with a barge pole.

NICK HILLIARD: It's pretty poisonous, that's for sure.

REMCO: And I really applaud your efforts in trying to ensure fairness and transparency in our policies. What stumps me, and if you have an answer to that or the chance of an answer to that I am more than happy to hear it, but I completely fail to see how any proposal to this ?? I mean, in this direction is going to get consensus in any extent. I mean, we are going to see people saying yes, and people saying no, and I'm stumped. I don't see how we are going push this through. I do see that we need to have this discussion. But I think in order to get this fairness in place, we probably need to change the policy development process itself.

Second, and that's just my gut feeling as a private individual. The one thing I am more concerned about than anything else regardless where the address space goes is that we use it as effectively as possible, which is ?? and that goes not for just private individuals or ISPs, but for the entire Internet. And I don't see how giving a block of address space to a single enterprise, whereas you could have used that on an aggregated scale with carrier grade Nats or some other nasty means to serve a larger community, how that's going to help anyone

GERT DÖRING: And that's actually a point that I wanted to remind the group of, that the last /8 policy is that way as it is because it was thought that ISPs might want to hide their NAT 64 behind a few public IPv4 addresses. So, this is the reason why LIRs are treated differently because supposedly they have lots of users behind this. Yes I know that reality doesn't always do what we think in this room, just as an extra remark, hitting the same mark that Remco did.

NICK HILLIARD: You can certainly argue that the address spaces are going to be spread out more thinly, but more thinly doesn't necessarily correspond to the requirements of the business needs of the organisation.

Yeah... there is no such thing as fair, we're at the bottom of the barrel. It's a real problem. I'd like to talk just briefly about the consensus issue, Remco, I think that's important.

The first thing is that consensus doesn't imply unanimity. Certainly we are not going to get unanimity on any proposal because we're effectively all one way or another, we're fighting for our little share of the last bits of address space, and unanimity is simply not going to be possible.

Whether consensus can be achieved, I am actually cautiously optimistic. You know, I think that there is actually a possibility of creating a proposal which balances the requirements of the LIRs with the requirements of the community which require provider independent address space. I think it can be achieved, I think we're going to need some sort of discriminators in there maybe in terms of pricing in order to stop the sort of potential gratuitous abuse which is likely to happen if we go ahead with the policy as it is. But I don't think I am as pessimistic about it as you are. Which is why I am up here and you're not. Okay, next question.

AUDIENCE SPEAKER: Hans Petter. I think we are facing looking for the root cause for this and maybe we are just fixing the symptom here and not the root cause. I think we're touching two different things one is do you want to become an LIR and a member and pay your fair share of running the RIPE NCC and this community? Or do you want to just pay the sort of small bucks for the address space and then you can still be part of the community and take part in these meetings and influence the process but you don't pay your share of the bill? I mean, that's one aspect. Part of the problem here is that policy is coupled with the size of the bill. So, from my experience, I would say that the best way to ensure your investment in address space is to become an LIR because then you have some certainty in what's going on and you can influence it by being part of the RIPE NCC and so on. That's my personal opinion on that.

Now, this is the sort of the one problem which is not really policy related but influences the policy. The other thing is we have this distinction between provider aggregatable and provider independent, and I understand that from a technical point of view that we want to conserve the memory in the routing and keep the routing table small. What I don't understand is how this concept is going to live in the next 10, 15 years when we don't have any free address pool, because then we are going to trade smaller and smaller blocks from existing blocks based on who is going to do what. So what we really need is a completely new framework where addresses where it can be transferred in a traceable way and where, on the other side, referring to my first argument, people contribute to financing the system that needs to do this. But that's a much bigger rewrite than just this. So...

NICK HILLIARD: Yeah, you're absolutely right. I mean, there has been talk over several years of collapsing PI and PA address space.

With regard to the problem that you first brought up, I think that's a RIPE NCC problem rather than a RIPE community problem. The RIPE community institutes 700 million people. The RIPE NCC institutes whatever number it is, 8,000, 8500 RIPE NCC members the certainly there is a huge disparity between the amount that you pay for becoming an LIR and the amount that you pay for getting PI address space. I don't have any easy solutions for that. It would be difficult to go back to 51,000 PI address holders and sort of say, well, look, for the reasons we have decided that we're just going to like jack?up prices by 100% or 500%, whatever it is, because we feel like it, that's a very difficult argument to make, you know, and there is a very large constituency of PI address holders out there and it's likely that there would be a substantial level of revolt if that sort of proposal were to be made because effectively the RIPE NCC is obliged to implement the policies that the RIPE community requests.

So, yeah... this is a difficult problem. Maybe the way to deal with it is to create the possibility of ?? to create the category of PI addresses from the remaining pool and deal with it that way. It's a bit messier, it creates yet another colour of IP addresses, but it's certainly one way of dealing with it. You are right, it's a bit of a mess. With no matter what way we go about, it there are going to be problems.

AUDIENCE SPEAKER: I'd like to make a couple of suggestions to maybe help you to move this topic forward.

First would be very nice to have as part of the presentation a data point representing how many reclaimed address space has been actually reclaimed say in last year.

The second one is to have information about how many LIR assignments have been made from last remaining /8.

And the third point is how many requests have been received by, in this case, RIPE NCC, for parks aggregatable address space versus attempts to get provider independent address space and that would be some data points which would help to move discussions.

And another comment is maybe you can try to look at separating handling of last remaining /8 versus that's been reclaimed. And possibly consider having a policy where PI allocations from reclaimed address space would be possible.

NICK HILLIARD: That has been suggested essentially, yeah, the proposal is that we create two new different categories. One is which is the last /8, which is 185 /8 and this other category of address space is reclaimed address space.

It's a possibility. It's messy. It doesn't really cope with what's going to happen when the last ?? when the real last /8, 185 /8 is fully decompleted by the LIR community, because at that stage, there is going to have been probably quite a lot of PI assignments from the reclaimed space, and we don't know what's going to half happen with that down the road.

The other suggestions that you made in terms of getting numbers, I think these are interesting numbers to have, but I don't think they necessarily change the philosophy behind the policy. It's merely getting a bigger understanding of the problem, it's not actually dealing with the philosophical problem and practical problem of look, well, do we have PI address space or do we not have PI address space? Because right now, we don't have PI address space. So, it's interesting but not directly relevant to the policy proposal.

AUDIENCE SPEAKER: About the last part, I honestly disagree because I believe that should be a data driven discussion but anyway.

NICK HILLIARD: Several of the figures were actually provided ??

CHAIR: May I interrupt for a moment. Alex is looks at me. I think he might have some of the answers already. Do you have some of the numbers right at hand?

NICK HILLIARD: Alex presented several of these figures in the first half of this presentation, which unfortunately clashed with the IPv6 Working Group and there were rather fewer people in the room so, Alex.

ALEX DE HEUX: I have a few rough numbers. Over the last year, we have reclaimed about 8 hundred thousand IP addresses number and PI space that we hand out is about 2 and a half million addresses per year, so we hand out three times more PI space than we reclaim. And over the last year, we have reclaimed space but now that we have reached the last /8, it's very well possible that we're going to reclaim a lot less space because in the cases where it would be available for reclamation, it might have been transferred away already during bankruptcy proceedings and that kind of thing.

CHAIR: Thank you.

AUDIENCE SPEAKER: It is the last space which we have and if we will give IP address only for LIR, then maybe we can get new IP space several years and if we will start to use this space for PI addresses, then I think in a very short time all the space will be used and next generation of ISP will not have any opportunity to get IPv4 addresses, isn't it?

NICK HILLIARD: I disagree. There is going to be an IP address market. This merely concerns the space which is handled by the LIRs ?? sorry, by the regional Internet registry rather than the addresses available on the commodity market. So, it will be possible after the last /8 has been depleted for people to go out and purchase IP address space from IP address brokers. But there will be a cost. There probably will be a higher cost than we have at the moment. So, there is going to be a possibility, but we don't know exactly what it's going to be.

AUDIENCE SPEAKER: I think money, it's very good way of regulation. If one amount of space is very low, then price will be higher. And what we give for PI addresses, then it will not work and I think many PI space will be good by some route company, for example we found several days ago that we have a problem with PI space, one person requested from different companies give for us registration document, sent a signed agreement and when we sent invoice to company, they called us and tell, we don't know this person, and I think if we will give for PI addresses, then amount of fraught wrong requests will be very high and PI space will be used much more faster than you think.

NICK HILLIARD: Yeah, I mean, you know... when the Titanic sank people ran to the life boats. It's tough. It's just this is the reality of life. I can't change that. Runout will happen.

CHAIR: So we do have renamed this Working Group to the Titanic Working Group, do we?

NICK HILLIARD: I like that. It was the No IP Address Space Working Group this morning, but the Titanic has an air of style about it that I like.

STENOGRAPHER: Don't forget the Titanic sank.

AUDIENCE SPEAKER: I am wondering if there is a regulatory concern about the PI space because if anyone is allowed to register as an LIR, there is no restrictions in doing that and if that's a prerequisite to actually getting address space in the final /8. Then there are no particular hurdles that we can overcome.

NICK HILLIARD: There are a couple of issues here. The first is that the RIPE NCC is a members club and the RIPE NCC has a very good working relationship with governments and governmental organisations and regulators, and it has been, it has always been in the position where it can go to them hand on heart and say we have implemented policies that are, you know, ground up, open and they are substantially fair. And if we sort of say, well, look, in the future, we have decided that actually we're going to keep this last chunk of pie, it's rhubarb pie by the way, it's very tasty, I love rhubarb ?? that that creates the possibility that they would maybe just sort of look behind the scenes and sort of decide that maybe they are going to do an investigation.

Now, I don't know whether an investigation into the RIPE NCC would actually uncover any disturbing truth. I don't think it would. I mean I think that in general it's a very open organisation, I think it's well run and so forth. But for the first time notice history, the RIPE NCC would be coming to the attention of potential regulatory authorities and organisations like the European Commission as being doing something which at the face of it means it's desperately unfair because the LIRs get all the pie and the end users get none. It's the sort of thing that would result ?? that could happen as a result of a complaint from a disenfranchised end user who wanted IP address space.

Now, you're right, they could become an LIR and they could get the /22. It would probably be a bit of a waste of address space because, you know, they may only need a /24, they may need less or whatever it is. But, regulatory authorities tend to act, tend to start investigations and looking at problems like this on the basis of complaints from people who feel as if they have been disenfranchised.

AUDIENCE SPEAKER: Could I put it another way, the other alternative that an end user has is to get address space from an existing LIR for which they could get connectivity from and I guess it's probably against policy, but potentially they could get space from our RIRs, but it's not ?? the point I'm kind of trying to make is that RIPE is not going, this is my address space and I am restricting it to a group of people that have to have been members already, do you see what I mean? I don't know what that makes any difference but I'll just stop there.

AUDIENCE SPEAKER: A follow?up to that F what you are saying that not having provider independent space makes this a regulatory issue, how are we doing this in v6? Do we have provider independence there and how do we deal with it?

NICK HILLIARD: We have provider independent v6 address space, yeah. It's fine. There is no problem. You can just apply for it and get it.

AUDIENCE SPEAKER: A question from Tore for Alex. How many new LIRs have signed up since depletion and do you have any indication of this number is approximately the same as before or if the signup rate as sharply increased?

And Elliott from Cisco has: I don't see many lawyers in the room and would suggest that the matter be referred to counsel, but absent that opinion, I see no reason why one would consider the policy based on such a trait

ALEX LE HEUX: Since we reached the last /8 we have seen about 20 new members, which is approximately normal. But we have to keep in mind that the process to signed up as a member takes at very least about a week and usually a bit longer. So, that the people that have become a member since we reached the last /8, most of them actually signed up organ the process before we reached the last /8. So it's not clear yet.

CHAIR: Okay. Thanks. Two more comments and I'm going to wrap?up.

NICK HILLIARD: That's fine. Just to comment on what Alex said. He gave a lot of these figures in the Address Policy Working Group presentation that he made this morning, so I'd recommend to everybody in the room that they have a quick flick through that and check it out if they are interested in all of these numbers.

LORENZO COLITTI: I'd like to echo Ivan's comment. Okay. So, we have PI space. It's all gone, right. We have this other space, there is really, really a tiny little bit left, and the existing channels to get that space are to become an LIR, it's an open process, anyone can do it. It was the same as it was before, I don't see any anyone could claim that that is being unfair or discriminatory, it's just we are scraping the bottom of the barrel. I don't see why we need to change our policies just because some of the avenues that people used to have to get IP space that was there and isn't there any more aren't available.

NICK HILLIARD: Yeah... I mean this, I think, is probably an issue of perspective. I deal with plenty of organisations who feel that they have a legitimate business need for a small amount of IP address space which will give them ?? which will give them a number of things including business continuity in case the relationship with their upstream transit provider goes sour. It gives them the ability to provide a technically slightly better service for the Internet presence that they have and they view it as an important part of their business operation and business continuity. That's what I see from the people that I deal with. If they have to become an LIR, it increases their costs by a substantial amount. They get more address space than they need.

AUDIENCE SPEAKER: Jan, I have got a few comments about fairness. It's also about ?? fairness is also about the old PI holders and the new want to be PI holders get handled fairly in respect of time. So, if someone was just two weeks late, he sits now before the empty dish where there is nothing, that's the one thing.

The second thing about fairness is if you say €50 is not enough to pay for a /24 nowadays because it's a really scary source, why not give it out for a quarter because you get a quarter space, a /24 and you only need a /24, why get a /22? But that's no reason to pay only €50, just pay a quarter of the LIR amount of money and then we could somehow find there a middle space.

NICK HILLIARD: All right. Three points in order.

You need to talk to a child about what is fair, because a child will always tell you that if they get what they want, that's fair. But if they don't get what they want, that's really unfair and they will scream and yell and I think we are all children at heart.

The second issue is about do we charge a quarter of the amount of what it would cost to operate an LIR? And essentially, that is a policy issue for the RIPE NCC to deal with. I think there is a very strong body of opinion in the RIPE NCC Board of directors, and I could be open to correction on this, that they would feel very strongly that they are not charging for address space as a commodity, which has a particular price tag, and if we create a policy which says that, well, look, you can get, say, a /22 for this amount of money, but if you want to pay a quarter of that amount of money you'll get a /24, well that assigns a specific amount of value to a specific amount of address space, and my understanding is that this may affect the RIPE NCC's relationship with the tax authorities. I see Remco jumping up there but hang on a moment Remco, one more issue.

Which was the issue of actually the cost of a PI address space which is ?? sorry the charge for PI address space, which is €50. And again I am open to correction on this but I rather suspect that the €50 per assignment that the RIPE NCC receives probably doesn't even go a fraction of the way words to the cost of actually providing the registration services and implementing proposal 2007?01 which ultimately needs to be amortized over several years. So that's probably a question that Remco or Axel would be able to answer better but I suspect that yeah, there is a huge disparity between the cost and the price. This is a policy issue that we're dealing with here. This is RIPE community. The costings are RIPE NCC. It is handled separately for good reasons.

AUDIENCE SPEAKER: Actually, it was kind of a follow?up comment to Hans Petter, who said that the amount you pay for PI is much too low, so...

CHAIR: Okay. Please people we are ten minutes behind schedule, so one very, very last comment and one very, very very last comment.

AUDIENCE SPEAKER: The other issue was abuse of PI space if we should find a way to assign it anyway. One way to deal with abuse would simply be to make the restrictions harder to get PI space. So, I could imagine that you say, well, if you are an end user and you obviously want PI space you want it for dual homing, okay but you not only have to say you want to do this, but you have to sit on PA space already for quite sometime, and you now want to switch to get multihomed. That would be a reason, and there wouldn't be so many end users fitting into that category. So, it's just to find a way to find something in between the two positions, don't do it at all or do it as you have done before, so, find some way in between. I do not know whether that would be the right kind of restriction or if there are other restrictions you could impose on that, but if we are looking for a middle way, we have to somehow find something.

NICK HILLIARD: This is a difficult problem. I suggest if you haven't read it already, there is an excellent book called Gödel, Escher, Bach, which was written by a gentleman called Douglas Hofstadter. It is a deeply philosophical book but it deals with exactly this problem of, you know, can you create a record which cannot be played on a particular record player, and essentially, what you are proposing, and to an extent what I am proposing is, well, can we create a set of rules which cannot be abused? And the answer to that is flat no. We might be able to reduce the abuse but we can't stop it. And on your second point, I'd like to reiterate what else Neilus said earlier. This policy isn't about punishing those people with a legitimate need for IP address space. So, yes, thank you very much for your comments.

CHAIR: Very, very last comment.

AUDIENCE SPEAKER: Remco can. I have got three seconds. Remco, as the Treasurer of the RIPE NCC, I couldn't have said it better Nick. What you conjectured about what the opinion of the Board would be. If somebody else wants to put an price tag on an IP address, they are very welcome to do so. I am not touching that subject. Thank you.

CHAIR: Thank you very much. Thank you, so Nick, thank you very much for bringing up this touchy subject.

NICK HILLIARD: I think that's a record.

CHAIR: And tasty subject by the way.

NICK HILLIARD: That's a record, five minutes of talk and 45 minutes of discussion, I think that's a record.

CHAIR: That was expected. To wrap it up, what I have heard from the room was very clear yes we see the problem, but I also heard a very, very clear that the proposal on the table is not what we want. So, I think the message to Nick is pretty clear. Digest it and try to come up with a new version of the proposal that might come somewhat nearer to consensus and then see what happens with that. It's going to take a while to work on that. But thanks for trying. And let's see how it goes. It will go to the list anyway.

So now the next one is actually a quarter of an hour delayed. Let's see if Milton is still here. This is 2012?05 the transparency in address block transfers.

Calling right now. Hello Milton.

MILTON: Hello.

CHAIR: Sorry for the delay, we actually had a lively discussion on IPv4 PI space before that. And now we're at the transparency of address block transfers. I put up a slide that will show the text in the policy proposal so people can see it and read it and we'll just give the microphone to you to, well say something about it. I would like to show your face to the crowd, but then they cannot see the policy proposal.

Well, let me just briefly recap it. The NCC is to document in a public accessible form any transfer that takes place with the name of the seller, the name of the Receiver and so on, and it has been on the list, feedback has been ?? there has not been that much feedback but one of the things that have been asked for was that rejected transfers get published original in anonymised form because that's the way the NCC usually done rejected things. And that's all from me, now you get to discuss this with Milton. Thanks for getting up early and being here with us.

MILTON: Can you hear me okay? So, all right, it's a very simple proposal basically. As you know we had a list published in the APNIC region and the ARIN region basically showing what transfers had taken place, and we felt, policy analysts and social scientist looking at this process felt it was very important to have information about the developing IPv4 market, we need to know whether resources are becoming concentrated, we need to know whether the market is working, we need to know how it's working what kind of resources are being exchanged, and it would simply be a lot of easier to know that if we published, kept track of the transfers that are taking place.

Obviously, if it were to become even more transparent it would be nice to know the prices but that seems to be too controversial to include in the proposal. And as we said, it would say also be good to know more about what is happening during the needs assessment process, and so it's good to know how many of these proposals are shot down or refused on the grounds that the buyer does not have the appropriate need for the resources.

So that's basically it. Again, I think it's a very simple proposal. It doesn't require a lot of change or a lot of activity. It's simply a mechanism for keeping track of how the market it working.

I am assuming also that you can see the proposal, that it's posted up there?

CHAIR: It's not yet visible.

MILTON: I wonder should we get rid of my face and put up the proposal?

CHAIR: I think if you just share the application on the screen and then it will make the face smaller and show the slides. But, basically, I don't think we need to see it to give you some feedback on it. I already see one person standing at the microphone. Hans Petter please.

AUDIENCE SPEAKER: Hans Petter. I think this is an excellent proposal and I support T I think we should have the openness and transparency in transfers so I think this is good. I understood you said that pricing is not part of the proposal, and while I can understand that, I think that maybe we should look into a way of doing some openness and transparency in pricing as well, so the same way it's done at stock exchanges. I'm not sure if you can see the individual trades but at least you can know the current day trade of that. So I think that should be the vision for this market as well.

MILTON: Yes, I think the pricing issue may be the next step where you have she is listing services could be, or maybe these services will arise through intermediaries and not directly from RIPE, but I think that the market would be a lot more efficient if indeed the prices information were known and, of course there are people who don't want the pricing information to be known and this proposal simply chose to sidestep that issue at this stage and simply make sure at least we know what transactions are happening more easily.

CHAIR: Actually, the pricing information might be a bit difficult to come by as the NCC, who is asked to publish this year, doesn't normally have the price. So...

MILTON: It may have to arise through a third party ??

GERT DOERING: I can see that we might want to have that eventually.

Any other comments on that? Support? Or position?

CHAIR: I think we exhausted them all with the PI discussion.

MILTON: I am sorry I wasn't here for that.

CHAIR: It was quite a lively discussion on whether we want to have PI from the last /8 or not and under which conditions. And which bites head.

Then I think we basically do the next round of the policy proposal on the mailing list with the change about the anonymising the rejected transfers, and then we can go forward.

Thank you for getting up early.

MILTON: Okay. Thank you.


CHAIR: We are at AOB anyway. So, any other business? No. Well then that's it for today. Thanks for being here. Thanks for contributing. Thanks for letting us know what you think about stuff. Enjoy your lunch. See you back in Dublin. And I hope for lively discussions on the mailing list because if nobody speaks up, proposals go nowhere, so please speak up, even if you think it's quite obvious that something should go forward, even then speak up. Thank you.