Live Transcription - Tuesday - 16:00 - 18:00
Note: Please be advised that this transcript is the output of the real-time captioning that was used on a trial basis during the RIPE 55 Meeting. In some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid to understanding the proceedings at the session, but it should not be treated as an authoritative record.
ROB: ... We thought let's have a little bit more focused time dedicated to these issues. It [will] not be a complete exhaustive overview of the [IPv4] completion. I want to remind you that in other [] area regions in the world, similar discussions [take place]. There are a couple of proposals floating around on what to do between now and then. Some of these have been submitted to the RIPE policy process and will be discussed, presented tomorrow in the first session in the morning, and will later be discussed in the address policy working group. So I don't think we should go into detailed discussions of these proposals this afternoon.
We have three speakers today: Mark McFadden, who, today, speaks on behalf of [] /ET nothing; Remco van Mook speaks on behalf of a RIPE participant on some ideas of what to do between now and then to make life after [] 10 easier; and Geoff Huston who will point out why we are on the path to [] /TPHRAOEGS.
So without further ado, Mark, if you could start.
MARK McFADDEN: Hi, my name is Mark McFadden. My day job is with BT, but for this presentation I am from the European Talk Communications Network Operators Group. And that's an unusual thing []. I [thank] Rob and the organisers of the Plenary for giving us a little bit of time here. Often, people who come to the Plenary are representing a single organisation or themselves as participants in RIPE, but this is sort of a special thing for our group in that the telecommunications network operators in Europe have thought very carefully about IPv4 exhaustion, talked about what it means to them, [] /PWAUBGS that it means to the [] /HREURBGS and RIR community and wants to [] sort share those talks with the community at large.
One of the things I am looking forward to hear [] /SKWRES talk, one of the things that's sort of interesting at this point we /THUPBG that the exhaustion issue is fairly well understood, that the IPv4 free pool is gradually [], and, to be honest, quickly be exhausted is something that there is very little contention on. It's whether you use [] pal /TPHOEPL knell or [] ex /POEPB he thinks models really doesn't matter. The date actually shifts by a matter of months.
So the actual issue of exhaustion isn't, well, in question, and Rob did a good job here starting this part of the Plenary. The real question is, what to do when that moment comes and what to do [] it the transition period? That's a really good way to describe the problem. It's very important, as you can imagine, to the text network operators who, in many countries, are the largest ISP in those countries, is that that's interesting that that group has a subcommittee that actually works on IP addressing issues and they actually built a consensus position that I am going to describe in a few minutes. The idea here is not to rehash the exhaustion model. Like most of us here, I am looking forward to Geoff's presentation, but when I actually see the curves, what I started to think to myself is, well, where the curve intersects 214 or 222, or whatever number you want to use, that doesn't matter so much as what happens in between now and then and what kinds of policy proposals, what kinds of thinking should go on between now and then in order to get ready for the time that the free pool is exhausted.
One of the things that we know, I was in [] /AL better Kirkey at the recent ARIN meeting and there was rich and well informed discussion on this topic, is that there was just no global consensus on what to do in terms of transition. Here at this meeting in the address policy working group there will be a discussion on a variety of proposals. There was a variety of proposals presented at ARIN last week and when LACNIC takes this up and [] happen nick takes this and AfriNIC takes this up, there will be a variety of proposals as well.
There isn't even an agreement on a combination of proposals and that's one of the things that we want to bring, sort of, your attention to here.
What there is, is a lot of well informed discussion on the issues and policy proposals that are starting to emerge, some in individual RIRs and some where the proposers of the policies are going to the [ ]all the RIRs, trying to produce a global [proposal].
While all this talk goes on, what we want to remind [you is] that the free pool continues to get used up. It's good to have informed discussion. It's good that we are, as engineers and operators of networks, we are aware of the fact that the global free pool is drying up, but it continues to dry up as we talk.
What ETNO did, and just on in my 15 seconds on what ETNO is, is that's it's the European Telecommunications Network Operator, so it's the network operators themselves. It's about 44 members, I think, around Europe, have a group of experts that basically a subcommittee it's a working group, on numbering, addressing and naming issues. And that group built the consensus policy that you are about to see, and it's a consensus paper that describes what good policy would look like for this period of transition, and while we are using up the rest of the free pool. This last bullet is very important to me because it's a real difference from the way that we work in some of our other communities. We are so used to the words "rough concensus" and we are so used to the idea of consensus building being sort of a community idea where maybe not everyone has to buy in, but if almost everyone buys in, then we have a standard or we have a policy.
In ETNO, the situation is quite a bit different. All 100 percent of the voting members have to agree to a position and, in fact, this one is, when it's called a common position like this, this is an example of a position that all the members of the ETNO have agreed to.
I am going to talk about the principles, and I was talking to [] GAR in the break. What we have here is a set of five principles that are in the common position, and one way to think of these five principles is that they reflect what good policy would look like in this period of transition while we are using up the remaining free pool. What would good policy reflect? So, what you have before you is actually not a formal policy [proposal] in and off itself, but perhaps a means to help evaluate what a good policy [proposal] would be.
I am sure many of the folks in the room would agree that there is no need, at this time, to abandon the existing RIR processes. A key principle here is that the RIR processes work. The policy development processes that are the bottom up processes by which all five RIRs use to build their policies and we can actually coordinate them globally, come up with global policies [] is working, there is no need to change it, there is no need to work elsewhere for mechanisms to address the issue of free pool exhaustion.
Very, very specifically, the second bullet is very important here. This community of network operators does not want to see the intervention of other organisations in this process. The RIRs do an excellent job in the communities that support the RIRs, do do an excellent job of building the policy and managing the numbering resources that are so important to us.
Also, and this will be important when I talk about market places, ETNO doesn't want to see Government intervention in this. I don't want to [] you'lley look to individual governments or groups of governments, if you will, [] reg /AL government, to help me divvy up the IP address space. Intervention of governments is just going to be a bad thing here.
And so, once again, very, very importantly here, whatever policy is built to do, to help us manage this transition period, it should be policy that's built in the common bottom up way that we are so used to amongst the RIRs.
There are a bunch of proposals for IPv4 exhaustion that suggest that there ought to be set asides or countdown periods. And ETNO doesn't believe that those are necessary; that that's an example of simply pushing the date of exhaustion closer to us. So if we set aside N, and let N be anything you want, whether it's 5, 25, there are many proposals out there. What you are doing is simply accelerating the problem, making it worse for everyone, and so what RIPE should see and do, in fact what all the RIRs should do, is use the allocation policies that are in place. There is no good technical reason for doing set asides only. Political realities there.
After yesterday afternoon, this was the slide I am going to spend the most time on. ETNO believes that there should be no IP market place, and while I have talked to people about this particular slide, some people have suggested that this shows that ETNO has its head in the sand. I am going to claim otherwise and I am going to try and convince you that an IP market place is actually a bad idea for us.
First of all, an IP market place demands that there is some liquidity in the goods that are on offer in the market place, and ETNO believes that there would be no liquidity or only a small very modest amount that would not affect the lifetime of a free pool in any meaningful way.
Also, a market place is probably contrary to the principles of fair [] place and conservation that underlie what RIPE does. If you have real needs for IPv4 addresses, you should be able to get them. In the market place where available capital was what dictated, not need, but available capital, fair play might not be a feature of that.
Also, ETNO thinks that RIPE itself shouldn't be in the business of actually being the market, shouldn't be the market maker, it shouldn't actually be the market itself, and one of the things that RIPE could do here, in our judgement, is actually, through its certification programme for the blocks, actually discourage trading. Now, trading is going to happen, let me just straighten up and ['fes] up about this, is that there are going to be some examples of moving IP address blocks from one organisation to another. But reflect on this: RIPE already has rules for that, RIPE already has established policy for address block transfer. And so let those rules work. Once again, don't abandon the RIR processes here; let them work the way they are supposed to. Other people suggest that there is the potential for enormous black markets available here. What ETNO feels is that there isn't enough liquidity there to make a difference, and so, go ahead and discourage the folks who would do trading illegally, right. Use the certification programme as a mechanism to do effective management of IP address block transfers, for instance, in mergers and acquisitions, but what RIPE should do is not encourage the appearance of a market place, because the appearance of a market place, in the end, is going to attract regulatory attention. In the end, there are going to be competition issues, there are going to be antitrust issues and all the things that regulators like to look at in terms of the fairness of the market itself. And if, as we have seen in some of the proposals, market place is actually designed where the participants don't have voluntary restrictions but the restrictions are placed on home for what they can trade, how often they can trade and what they can trade, then you have the competition issues that are related to them, and surely, in those cases, regulators are going to be very interested.
As a result, our community really strongly thinks that the market place is not something we should encourage; that's something we should stay away from.
Instead, what we need here is a different principle, the principle that we have worked on all along, and that is that the allocations from the existing free pool, that's left of it, should be based on needs, as we have always done. Okay. Don't change that.
Geographical, regional set asides are an artificial way to actually accelerate the end of the free pool, and so we are against that as well.
If we actually do that set aside, what we are going to see is a global IP address shopping, that's going to actually that's going to be something that other communities that have available pool where we might not have it in Europe, what we will do is have global organisations, perhaps third parties, going around and shopping for those addresses.
Once again, if we see that kind of global address shopping, we are going to see governmental agencies interested in intervening here to protect their national or sovereign interests.
Another feature of policy is that the existing the existing RIR policy mechanisms work just fine. If new IPv4 policy emerges, let it emerge through the existing policy development processes that each of the RIRs have. In fact, another way to say this is the RIR community should do nothing special here. What it should do is rely on the policy development processes that already has in place and each RIR has formal set policy development processes that we think works. The ISP and carrier community needs one thing very importantly in this process of transition and that is predictability. If we start actually allowing policy to be developed in ways that are alternatives to the way that the policy development process works, we risk that predictability.
Another thing, and it was said by another member of this community and I can only expound on it here and say how important it is, is that as people start to think the sky is falling, they'll tell it to other people. And so what there needs to be is a more informed, a more broadly informed community about what's actually happening with the free pool. I think that the engineering and addressing community actually knows what's going on here, but I doubt that the BBC does, and the reason I doubt that is because I have talked to them.
Here is the deal: My mum doesn't know about address deletion, and frankly she doesn't care, but when she reads in the paper that the Internet is going to break because there are no IPv4 addresses, she'll give me a call.
What we need to do is make sure that a broader community is informed, a broader community than simply the engineering community, a broader community than simply the Internet policy one, a broader community than simply the people who are interested in addressing policy. And we, as a group, are the ones responsible for that. Right now, through the NRO, we [] /PWHREURB statistics when Geoff actually does his modelling and when we [] are people do the alternative models, they actually look at the statistics that the [], risks [] R has /PWHREURBD together jointly and publicly through the NRO. That's great. But nobody builds that information in a way that the BBC can understand. When the sky is falling, then what's going to happen is the capital markets are going to want to intervene. They are going to sense that our industry has a significant problem here. And while we do, we don't want them to have the 'Chicken Little' effect.
Also, what I'd like to suggest to the RIR community is that the work that Geoff Huston has done has been incredibly valuable during this period. It's worth noting there have been many other models they are not as well known as Geoff's many other models that basically have the same results. That's why I think the issue itself is not in question. What might be nice here is an independent model, a third party model that the NRO or the RIRs, as a group, commission, and so that's a suggestion of ETNO, as well.
There has been lots of talk about legacy blocks and doing retrieval of legacy blocks, and the ETNO community believes that that is something that is a mission of the RIRs. It's something that the RIRs should continue to do but not spend a huge amount of resources on. If people want to offer blocks back, let's make it easy for people to do that, but let's not consume. Leo had a good presentation yesterday on what it took to get [] net 14 back. And here is the deal: it's too much work. So the RIRs, as part of the principles of conservation that they already have, should make it possible for people to return legacy blocks if they want to, okay? But actively, should they go out and do active work in retrieving legacy blocks? I think not. I think that what we have seen already is the amount of time that would give us for the transition. A single [] slash 8 doesn't give us much time at all in terms of the runway.
Now, ETNO, as an organisation, it's great for me to represent them up here because, as an organisation, what you really have is sort of 43 people up here in a way, but, speaking for myself, one of the things is that I'll be there in the address policy working group tomorrow or on Thursday, I am sorry, and one of the things to see about this is that we don't have, on the table here, a formal policy proposal. Instead, for us, what our common position paper is sort of a five guidelines by which we decide where we can actually help inform the dialogue about what would make a good proposal. And those five those five guidelines, if you will, are right here.
The existing IP addressing community is the place if any works needs to get done, let's do it here. Let's not go someplace else, a body that that is [] annually letter or five letter acronym.
Allocations, now and always, should be needs based. That's based on the fair use and needs based that all [RIRs] have been built on since the beginning.
We want to strongly discourage the emergence of an addressing [market] place because we think it invites regulation and government activity into the IP addressing community that has always been self regulated and has been effectively self regulated.
The existing policy development processes should always be used and then the legacy blocks themselves are an activity that ought to be treated separately from the transition here during the time that we have until the end of the available free pool.
That's it, Rob, and I'll take questions.
SPEAKER: Start queuing up for the microphones. I'll give the floor to myself first. And not a question, [] and remark taken by one of your last slides where you basically said that we all know about this problem, but this great world outside who doesn't know about it or, even worse, have wrong impressions of the problem.
We have, this week, activity, I think it's in again in the address policy working group, where we will discuss a possible statement of this community covering this, which is meant not so much for the community internally, but for the rest of the interested environment, and the idea there is to basically what you described here, to say, yeah, we are aware of this, yeah, this is going to happen and we are working on a smooth next three or four years' operations, and if you are interested, come and join us. And, so, it might not be exactly what ETNO had in mind, but at least it falls into the same spirit.
So if you are interested in formulating that statement, I think it's in it's on the [agenda] of the address policy working group.
[MAN:] This is exactly why I am standing up here. In the policy working group, we have two working group slots this week. The Thursday slot is dedicated to discussing end of IPv4 proposals and starting to discuss [] while the community resolution on IPv4 exhaustion and the discussion will then go onto the IPv6 working group on Thursday afternoon.
[ROB:] I'll take this microphone, then that one and then that one.
[SPEAKER:] Hello, Mark, you have quite accurately described, I have to say, I agree with most of it[,] what ETNO would like to have and what they would not like to have, but after we have run out, what does ETNO think that we can if we are going to needs based calcations, where does that come from? What idea does ETNO have about redistribution in that case?
MARK McFADDEN: Two questions there. There is the question about redistribution and what happens when the free pool is gone.
I sort of talked a little too glibly about the day the free pool is gone [because there is a] period in which IANA has no free pool. It's [] /TKR BGD to the RIRs and there is the period which the RIRs have no free pool to [distribute]. But one of the periods that happens in that period is we also have people who are being pushed to solutions that we may not like as engineers, sort of [] net PT solutions, and so forth, and then a solution that we do like as engineers towards IPv6, just for the addressing part, not for anything else IPv6 would have. In terms, actually, [] /TR FRG address, I think RIPE already has policy on the books for what will happen in cases, for instance, of mergers and acquisitions. And if you here is a place where you and I probably disagree, but if you imagine that there will be companies that come along and acquire other companies just for their IP address space as assets, imagine what governments might think of that, right? Well, if that really is an asset, if an organisation is bought for that, then perhaps we should tax them on their assets. And then what would happen to organisations that had, you know, a [] slash 14 here, a slash 8 there? So, in terms of redistribution, I think ETNO thinks use RIPE's policy that we have already got. In terms of what happens in terms of allocating, what do you allocate when the available pool is gone? There is nothing to allocate, as far as I am concerned; either you are doing things [that are] unacceptable to us as engineers or you are moving, you are getting on with the transition.
SPEAKER: Okay, but the point I am getting at is it's very nice to have policy, but if you don't have any addresses to apply that policy to, well, then...
MARK McFADDEN: I understand what you are saying.
[NEXT QUESTION]: Hi, Mark. I agree with many things of what you said. Of course, I agree [with] the commitment that you express on behalf of ETNO regarding the current systems, also the principles that you have stated.
I have one point, is regarding the market. I, personally and this is a very personal [] continue I, personally, think that we cannot discuss about the market very much at this stage, because for discussing about market [] /PWAEUFRS we need to know exactly what the market is, and I think at this moment we cannot know exactly what the market will be, if there is, because it depends, among other things, on decisions that could be taken as the policy that you have mentioned before. So I don't necessarily disagree with what you say regarding the market, but I disagree with the fact of making that sort of statement at this point because I think we have to remain open minded and come back to this issue later, probably next year.
I have two disagreements with what you say. One of them is regarding your pessimistic view about the policy proposals that are being discussed. I am much more optimistic. I think that after four RIR meetings, I am optimistic with regard to some of the proposals that are being discussed will be approved, but also that there will be a series of proposals that will be seen and approved in the next one or two years.
My other strong disagreement is, you said that do we approve of some policies could mean inviting the governmental intervention? And I think exactly the opposite. I think that doing nothing is inviting the governmental intervention. Thank you.
MARK McFADDEN: Maybe I could just respond to one of the things. I am [glad] that you brought those up.
The one that I would the remark I would make is for those people who think, or who have the point of view that a marketplace might be inevitable or might be a good thing, one of the things to consider is what effect the marketplace would have. Ask yourself, well, if a marketplace emerged, would it have any effect on elongating the period of transition to the end of the free pool? And I think one of the comments there is that we don't know the answer to that either. And yet, the marketplace opens up other dangers. One of the things there is that a marketplace asks for or requires liquidity in the resource that's being bought and sold, and there is no guarantee that liquidity will exist in an IP marketplace.
QUESTION: Shane [Curr] from ISC. I think that, based on your answer to the first question, that, basically, ETNO says that there is no plan, once the addresses run out, to change anything. So basically, we don't do anything until the addresses are gone, and then nothing. That's the plan, right?
ANSWER: I think I would probably put it a little differently. I won't even and while I admire the spirit in which the question was offered, I guess what I would say to you is, look at the other proposals that are in place, and they do the same thing. All they do is change the date at which that end point is actually reached. What they don't do is they don't provide any predictability about when that end point would be reached. So I guess what I am saying, and I think what ETNO is agreeing to here, is use the processes that are currently in place. Don't change them and don't make some special rules here. And what I would argue is each one of the other proposals has exactly the same problem. It just tries to change the date at which that problem would take place.
QUESTION: I think the difference that we are looking at is most of the other proposals well, certainly these ideas about marketplaces, admit that there is going to need to be some sort of mechanism to move IPv4 addresses around after we run out. I mean and I think ETNO is also an organisation of very well established existing people. They have got their addresses. To be honest, there is no real reason why they would want their competitors, which may not even exist yet, to suddenly be able to get these resources. So, if you don't do anything, we end up in a world where the people that were early adopters, the people that have their resources, now continue to have them, and the way the Internet works now, if I have a great idea and I start a new company, I can get the address resources that I need. After IPv4 is exhausted, if there is no mechanism in place to get those resources, those ideas, those companies, those organisations, those technologies simply can't be created. You can create them with [] /TR RBGS 6, but it's going to be a while before that's a viable option, really. Basically, I understand the strong desire to maintain stability. I think the way to do it of doing nothing is not really viable, because when the addresses run out, that stability is going to end, no matter what.
ROB: I think it's not fair to call it "doing nothing". I think what ETNO says is you have very solid, well understood, extremely workable procedures in place, don't abandon them.
MARK McFADDEN: That would be exactly what we say.
QUESTION: Cathy [Arnson], ARIN [Advisory] Council. I just have a quick question, and not much else. But one of the premises of some of the other proposals, for example, at some certain date you give each RIR a [] slash 8, is that it provides some predictability, because if you are not the guy who goes back right at the right moment and gets space, somebody else might get two [] slash 8s while you get nothing, and then there is none. So when you are talking about predictability, I think that that might provide more than the plan on the screen. So, I just thought maybe you could comment.
ANSWER: I think well, the first thing I'd say is I don't agree, but that's a really risky thing to say to you.
The second thing is that I think the set aside concept puts us in a situation where people who have global networks go shopping, and I come from a company that happen [] has a global network, and I don't want them to go shopping. I don't want them to go to LACNIC because LACNIC is the only place I can get a [] slash 12 from, and so in terms of predictability[,] I want to be able to use the rules, the needs based rules that we already have in place to provide the resources that I need. I know I shouldn't have disagreed with you.
QUESTION: The only comment I have to that, though, is if ARIN gets the last two slash 8s, then somebody is going to shop and get them anyway. I am not saying I agree with these proposals, but the premise is that everybody gets an equal amount at the end. There'll be less shopping.
ANSWER: I think then, basically, we just disagree on which is [the] more predictable end game.
ROB: And since we are talking about one of the specific proposals that says if there are only five slash 8s left, hand them out to all of the five RIRs, then it's over and done with, I think it is a minor detail in the whole complex of okay. Next question.
QUESTION: One of the authors of the proposals that [] being talk here, and I understand and it was not the intent on this gathering that this session to discuss the proposal [itself]. However, there has been some references to them. I invite everybody on Thursday to discuss what's being proposed, but I want to mention a point that was said, that there was not agreement in the different regions and I want to say that, in several regions, there is already consensus on proposals, LACNIC, AfriNIC, we just came from ARIN where, also, there was consensus on working on this proposal. So I think I disagree on the comment that was made.
ANSWER: Let me just be clear on it. I was in [] /AL better Kirkey. What I didn't [] was consensus on a particular proposal. What I heard was consensus to work on proposals.
QUESTION: That's what I just said.
ANSWER: I believe there is a a global working on proposals for a global [] /WOPBS.
[WOMAN]: Can I [ ]just a point of clarification. I am an ARIN Advisory Council Chair. There was two proposals [globally] and we just abandoned one only so that we can could take the authors from that one and combine them with the authors of the other one. So we are continuing to do work but we want to combine it all.
[MAN:] We are working on that. The other author is also here. What I mean is nobody knows, nobody has an idea. We are working on those.
ROB: Okay. Right. I think we should have three more short and quick questions and then we go to the next speaker. After the last speaker, there should be ample time left for not endless, but for an interesting discussion. So, if each of you can be short and to the point.
QUESTION: Todd Underwood, [] reNaas us. I take your point that a market has some unknowns, so you say Geoff Huston's work on the time frame, in one of his earlier papers he suggested that we might get to 2030 if a market was allowed to continue trading under various models. So if you buy his models now, which look awfully good, do you just not accept that last part of it? Because 2030 is sounding great to me, given what crap IPv6 is. I mean, 2030 sounds like we can throw that garbage away and get something that works. Wouldn't that be cool? 2030.
ANSWER: I think Geoff in [] /AL better can you recollect eye [] that the dates you get depended on the assumptions you build into the model, and so I think that the assumptions built into that model are open to question.
QUESTION: But the point I want to try to make, so you just don't believe that an address taking market substantially extends the life or extends the life of v4 usability in a quantity sufficient to outweigh the unknowns that it creates.
ANSWER: That is very eloquently said.
QUESTION: Niall Murphy, Google.
I actually agree with you on a lot of things. I think we should try and preserve the good qualities that we have in the existing system, absolutely. I just think that we are sort of in a state of denial in that we don't understand the urgency of the problems, particularly for new entrants at the time of exhaustion on [] /SELT. I haven't seen anything in those slides that addresses that question.
ANSWER: I take that point very seriously, and there was another speaker before you who brought up a similar point,[] is that what happens to new entrants in this, and I think that's a question that this group has to address. Fair enough.
QUESTION: I have a short question, but I will try but the question is: What we can do to accelerate IPv6 deployment, because is there any position on that?
ANSWER: There is among its members, but one of the things that happens among the individual members is that the business dynamics of individual companies are different in terms of how they invest in IPv6, and I think just speaking for my own company I think what happens is as more and more awareness comes into the engineering the engineering community, there is more and more awareness of what the problem is going to be for product development and roll out of new things to consumers. And since that's where the revenue is, what this is doing is actually having an effect on the revenue of organisations like this and that's a very, very meaningful way to get IPv6 investment.
ROB: Okay, Mark. Thank you very much for a very interesting presentation and thank ETNO
(Applause)
thank ETNO for formulating a very clear position.
The next speaker this afternoon is Remco van Mook, who comes well prepared, I think.
REMCO VAN MOOK: I have come well prepared, as you can see. It's not made of [] calf lone, so anything suffered and rotten [] flute, please, because I need to make it to that exit.
That being said, what I am going to present here is pretty much along the lines that's been discussed worldwide over the past few months. I have given it a soothing subtitle, perhaps we can work something out, and if we can work this out, we can do world peace in an afternoon, I think.
I don't know what this is. Okay. So my company is sponsoring RIPE this time around, has nothing to do with this presentation. There is an on line survey. If you answer, you can win one of five iPod Shuffles, and they are all nice and they fit the general requirement for giving away [] venues like this. It's got an Apple logo on it. Go ahead, enjoy.
Here we go.
So in policy terms, what are we looking for? We need a great idea, and after we had a great idea, we need to execute it really well. So according to this graph, we need nothing short of a miracle, or a [] C.
What's the future of IPv4 going to look like? I have established this time line, but I didn't put any dates next to it, except for the "You are here," and as you can see at [ ]the right part of it is where it becomes really tricky coming to the last bit, gridlock, and we can establish gridlock when, as people call it, liquidity has run out or there are no other blocks available, what I am going to call it anyway at what price whatsoever. Or we have established that we are now running an IPv4 routing table, we have 16 million entries and things will have come to a complete stop.
And anything in between becomes well, it becomes more and more problematic to continue to run IPv4 on it. By "problem" I can I mean more costly.
So what happens here is IPv4, as soon as we go through the allocated pool, the better net will continue to work. It just becomes a bit more difficult for new entrants to participate in the Internet, especially if you are wanting to provide content to that Internet.
So, what's going to happen? We are going to run out of fresh IPv4 space soon. We have to try very hard that we don't run out faster than we have to. Running out of space, does it mean that we immediately lose the Internet? Contrary to some commonly held belief, running out of IPv4 means switching from analog to digital TV. Just over half the issued space, as you can see from Geoff's graphs, is actually routed on the Internet, and basically what we need to come up [with] is methods to use IPv4 space around ones we have run out.
Executive summary: What am I about to propose? We need to prevent further opportunistic claims on IPv4 address space, and I have tried to put it politely. I said, further, there might be some [] /TPH GSs that there have been opportunistic claims. You might look at your last allocation that you have got yourself and ask yourself whether it was completely genuine what you were seeing there. And RIRs will, from now on, be able to reallocate or designate. I have just a couple of terms to come up.
Allocated other [RIR] [ ]I am looking now if there is any rotten fruit here. Good, no.
Preventing opportunistic claims: We need to have a very careful look [at] what we are currently doing. We have currently set policy [] is enough to prevent people from coming in, make some outrageous claim which they can somehow fiddle through the current set policy, walk away with a huge chunk of address space and holding a valuable piece of property, whatever way you want to put it, over the next few years.
Current policy in principle has [] /TPHEF barriers, but that all depends on the quality of execution, so we have to look at that really carefully.
In the recent past, there has also been a couple of other suggestions. I have seen an aggressive slow start proposal which, in short, comes down to everybody who hadn't asked for new allocations in a certain period of time is going to be treated as a new [LIR], or a one way system, [](LIR) which means that once you have gone through alternative means of acquiring address space, you are not entitled to get new allocations any more.
IPv4 up to 2010. The current policy doesn't provide for an after market. We don't have anything in place which we have a few things in place, but they are rather fake, probably intentional, on how we can move address space around.
Reclaiming: This is not to disqualify the excellent work that Leo has done, but reclaiming is not going to keep up with the amount of space that we are consuming.
Access networks will eventually move to v6. Only content will have to pretty much run dual stack for much longer. The demand for IPv4 by far outstrips any possible new supply. Say if we were to come up with a completely brand new 32 IPv4 address space somehow, we'd be having this discussion again within the next ten years, because this is the rate that we are currently consuming network space.
Content providers: Content needs to be available to an audience as wide as possible. Why do content providers provide content? Because that's the way they make money.
It therefore needs to have direct access to IPv4. There are companies who have done presentations here in the past which basically said that every millisecond of round [] strip time they save towards the [] eye balances to our consumers brings them an extra million a month.
Where will the content company that starts in 2018, just whatever year, get its IPv4 address space from? Well, the basic question is: "John may /TPHARD [] Keane's demand creates its own supply." If you have written a big time business plan and you are planning to make a hundred million a year on some sort of content, you are going to get that address space, no matter what.
A grey or black market are considered undesirable. We do want to keep a few almost going on.
And within the current policy, the only way you can actually get new space legally or within policy is to go to HP or IBM, who have plenty of space around and sell servers and to get a block of IP addresses with those servers, and if you are buying enough servers, they will probably supply that to you.
Access providers:[] are what they currently hold most of the large v4 blocks. No big surprise there. Once they move to v6, there are tonnes of free address space around on v4 and I have done a quick survey about a half year ago and literally all of them said that they are not going to return that space. They'll not go on record saying that, but that's what's come back from them. So they are not going to return the space, no matter what free claim policy we adopt, if they are going to agree to a new reclaim policy at all.
Well, so, in the future, if we do want to move address space around or optimise usage of IPv4, we need to start moving around addresses as soon as we run out of fresh addresses to allocate and we want to keep records on that to [monitor] responsibility, so who owns what and [who] is responsible for what, if nothing else. And the RIRs provide an excellent position, already have an excellent position to act as a clearing house, so keep the records, if you will.
And they have done that so far, for the [] LIRs, and we could move that forward, only moving address space between LIRs.
Reallocation: Reallocation is moving allocated but unassigned address space from one LIR to another one[,] essentially what we are doing currently and what's already outlined in mergers and closure policies that we have, except that we are not merging or closing companies to do such. The difference is that we can now do this with the extensive involvement of the RIR, which is currently described in that policy. And you could use this piece of policy to reflect well, I am going to call it, reflect, say, buying space or buying a title to space or buying the right to use an allocation, whatever kind of legal issue you would like to put into that.
And designation is pretty much the same, the only difference that you don't get that address space on a permanent basis, but it's got a return by date.
So, whatever you agree with, the other LIR has a return by date on it, and that's a regular commercial contract and that return by date can then be simply enforced with regular contract law. We don't need to invent a new economy for this. And this task really reflects well taking on a lease.
What the RIR should then do is, after they have done the administrative task of moving the address around, is hand over a certificate for that address space so that everybody knows that this has legally been transferred. You can see that it has a "sales slips" if you want warranty.
So, what it boils down to is we are going to have a set of policies that will, at the same time, prevent boarding address space which, in turn, preventing running out faster than we need to.
We get a mechanism in place where moving an IPv4 space around, once we have run out of the allocated pool. We are going to make sure that whatever we do stays transparent and we don't get involved with any commercial negotiation between LIRs.
What's missing in the proposal is moving IP space to LIRs in different regions, which is a hairy thing to do. I think that's something that we can only do once we have established similar things in other regions and doesn't conflict with this policy if we want to do it later on. And it doesn't include regulation. And why is that? Because regulation also has there is a bad side of regulation. The bad side is that you have to enforce that regulation and you have to ask yourself if the RIR is in a position to enforce that regulation. So with this policy, we are creating a level playing field for everybody who actually requires free [Ipv4] address space. It's about as level as anything else in the world is. It probably facilitates most distribution models that we can come up with in the next three years and what we don't want to do is set any other barrier to moving that space around other than signing up with the clearing house, which is what we already do with becoming an LIR.
So what's that environment going to look like?
[] /S FL for LIRs, that doesn't mean that there can't be another policy for another group saying something else. This policy is specifically for LIRs, specifically for allocated but unassigned address space, specifically only for address space, in turn, that has been registered with the RIR, so it has been allocated by the RIR or its RIR space [], that that has been registered to the [RIR] risks. The RIR is only to provide the clearing house to do the bookkeeping. LIRs free to decide on what terms they are going to move this space around, and we might very well see IPv4 as a listed commodity, who knows?
And with that, I'd like to open the floor for questions.
ROB: For the sake of symmetry, Mark will now ask a question to Remco.
[MARK]: The question is about your last slide, about there not being any regulation. There is no regulation from inside, but the two questions I have is that if you think that this particular model would be successful, one of the things that you have to believe is that there would be sellers, that there would be people with address blocks that would come to the marketplace, and what I wonder about is what evidence do we have that those that there would be sellers in that marketplace?
And the second question I have, and then I promise I'll go sit down, is the regulatory side here. You say well, gee, everyone just signs up to be an LIR and as long as they are part of this marketplace, there is no other regulation. I think that that's probably not what would happen, is that the regulation would come from outside; that the marketplace would get regulated not inside, not a self regulating marketplace, but an externally regulated one. For instance, what would prevent sort of a private equity firm from buying up ISPs, getting rid of all of their customers and creating an artificial shortage? Those customers have to go somewhere else. There aren't enough IPv4 addresses. This model doesn't reflect the fact that the regulation would come from outside and not within.
ANSWER: Responding to your second statement first. I don't think that whatever we come up with, if we can avoid what you have just described. So I think we can talk about it for two more years, but not come up with a sufficient answer to that. And answering your first question, it's pretty well much accepted in general economic theory that if the price goes up, there will be bidders, there will be people who will offer their wares, and so it's just a matter of simple economics. So if there is no liquidity, then obviously the price is too low.
QUESTION: Niall Murphy, Google.
Two points. I guess the first point is that, in terms of establishing whether or not there will be liquidity of some kind, I guess the difference between the allocated and the advertised space would indicate that there is some shortfall there.
The second observation I have, more or less mirrors Mark's, in that I am thinking that this proposal is an attempt to define a mechanism which excludes all of the problematic bits that we don't understand by pushing them, devolving them. I don't think we'll be allowed to do that. I think we will have to bring it inside.
ANSWER: Okay. One of the main reasons why I [] am all dressed down this policy proposal is exactly because of that. My view [is ]that the hairier points of this will take another two years before we can agree on those. But getting the mechanisms in place to do a trade, market, whatever you want to call it, are going to take some serious time to implement, and if we wait for that, for year [] pro years, then it will almost certainly be too late.
QUESTION: It could be that this is a very useful first step, or maybe even only step, but that's for further discussion. Thank you.
QUESTION: Neil O'Reilly, University College Dublin, but not wearing any hats. This is just a question for clarification. You mentioned a temporary reassignment mechanism. Is that on a use it or lose it basis, or what's the how is that supposed to work? I didn't quite understand.
ANSWER: Okay the really simple thing is that if you are an LIR, you have got allocated but unassigned address space lying around which you don't think you'll need for the next five years but might need it afterwards, you still have a mechanism to get a return from that without actually losing that space forever. That's the general idea. Exactly the same as instead of selling your house, putting someone in it who rents it.
QUESTION: Okay. So somebody else can use it for head room on their way to v6, or something, but they have to give it back even if they do use it?
ANSWER: Sure, absolutely. You can renegotiate a contract renewal, whatever.
QUESTION: Keith Mitchell, IFC. I think one piece of the puzzle which you didn't mention which I think will need to be addressed if we go for this market, is the whole area of dispute resolution. I have one of the most idle volunteer jobs in the whole [RIPE] community, is that I am on the address resolution panel. In the past five or six years since we created this, there is one dispute. The one thing that came out of that, I think, was that the more legacy address space you deal with and as you start digging in more and more address space, all the good stuff gets used up, is that the records as to who exactly owns that are going to get more and more murky and I still feel that there are at some point, there is going to be a sudden ramp up in disputes over exactly who owns address space, and we need to prepare ourselves for that, I think, as part of this mechanism, unless there is never going to be any disputes, despite the fact there has only been one so far. I think that's a possibility in the future.
ANSWER: I think you are absolutely right. That's why I am saying in this proposal, it's only about address space that's already been properly registered with the RIR, so that where there is no current dispute on it. I agree that, absolutely, as soon as people see that they can get a return from the RIR space [they will] probably want to register it with an RIR. [](mark) we have got a process for that, and you are going to see more work in that process, I agree, absolutely.
QUESTION: Thank you. [Carston] [] chief err, with [Deutsch Telekom]. I just wonder, if we look a little bit further down the supply chain, we have seen proposals about how to reallocate address space on the LIR tier. If we look at the customer side, there might be the same thing going on there as well, that some users of IPv4 address space have more than well, they will use in a foreseeable amount of time and might exchange that for money, maybe, or give that away to other folks. Is that part of your policy proposal or will it be?
ANSWER: It's not part of the proposal, but I can very well imagine that, that in the commercial negotiation that's already going on between LIR and [Zen] customers, that this will become part of that, and if you are looking at redistribution of address space and if you are taking aggregation into consideration and the LIR that has issued, that address space would be ideally suited to mop up address space from their end customers, group it together, package it, make it a single larger block, that is now unused and do something with that.
QUESTION: So, that would, of course
ANSWER: It's not part of this policy, but I can very well see that happening.
QUESTION: Still, LIR would be involved, so there would be no direct, say, negotiation between users and customers of IPv4 address space?
ANSWER: That's pretty much the same. If I wanted to, say, sell 1,000 shares in [Deutsch Telekom], I'd not go to the Stock Exchange myself; I would have somebody who represents me sell those stocks for me or even have, with larger agencies, they most likely have a reserve of those stocks so they put it into their own reserves.
QUESTION: Okay. Thanks.
QUESTION: I am [] /RAOEPB as kin son, but I am speaking for myself. I make an observation, if people here ever want to do authentication of prefix advertisements, for example in BTB, or something like that, there are going to need to be records at the RIRs of which block is associated with which person or [] O, and so whether or not one endorses or doesn't endorse a hypothetical future marketplace, the practical thing is there are going to need to be mechanisms so that if [parties] choose to trade, sell, lease, whatever, address blocks, that the RIRs can update their records and so that we can have some reasonable [] prayer of that. If people don't carry about authenticating BGP prefix advertisements, then this might not matter.
ANSWER: Thank you.
ROB: Okay. Thank you, Remco.
(Applause)
The third and last presentation today is by Geoff Huston.
GEOFF HUSTON: Good afternoon again. Geoff Huston. I apologise for being up here so often. This one wasn't my fault, I didn't volunteer this talk; Mr. [] /REPB Dick put my name down, so here I am and I am going to talk a little bit about some numbers, particularly around the unallocated address space exhaustion times and then I am going to talk about the after life, what happens after that.
I talked about this at the last RIPE meeting and I used this slide, which I thought was kind of cool. Don't you like it? The apocalypse death. I was told this was overly dramatic and it was the wrong theme. So, after a little bit of investigation, I think I found the appropriate one for you.
So, where are we? As well as outside the railway station on the wrong side? Let's have a look at some of these charts showing exactly what's happening in the IPv4 address space and some consumption rates before we go into the after life, let's see at what time it looks likely when we are going to meet it.
Address space really sits in sort of two major pools that we are really, really interested in. The ITF has held aside the equivalent of 36 [] slash eights. So there are 250 slash eights as you might recall, 34 of them are in the ITV pool. That includes that little bit we talked about yesterday, right at the top, 240/4. I haven't used that as usable because there does seem to be some debate. The IANA pool of usable [lash out] is down to 44 as of today. (Mark) And [] right do you 175.922; in other words, a fair whack of them are out there somewhere. Where are they?
That's a bit interesting. 106 only, that's less than half of the [sways] you can see in the routing table. For those people who look at liquidity of the market, you find liquidity assigned but can't see them addresses. There are 46 [] slash eights, the equivalent thereof. There are literally thousands of the things, various ways and sizes. If you add up all the space that's not advertised in the routing table, you get to 266 slash eights, approximately one third of the address space we handed out, and it's not in the table. What is that space? No slide on it, but it's predominantly the old class Bs. Yes, I am so old I even remember that. Addresses between 128 and 191 make a fair deal of that, and also there is some space in the top of those old Cs, 192, 193, 194. Interestingly, since the RIRs kicked into action solidly in the mid nineties, the amount of space that was given out and is not advertised is remarkably small, so that, in some ways, the policies that we have used and the way that they have been applied have been phenomenally effective. We said this space is for use in public context. It is.
So that one third of the space is actually a lot of history and people say, yes, there might be some issues around legacy, there may well be, but certainly there is a lot of space out there, one way or another.
The RIR themselves have 22 [] slash eights. You are going, wow, that's a lot of [ ]kind of weird. There is actually two kinds of address space if you look at the IANA registry. There is this key word called "various". We did this so long ago we don't know what we did, but we'll just market various. [] to now that we are coming back to this we are looking [inside] various very, very hard and we are looking at [your] database, I say, very, very hard and we are trying to reconstruct history as well as we can. And as far as I can see, I can see, me personally, and I hope everyone agrees with me, there are around 22 slash 8s inside the RIR system, somewhere or other, which we'll probably have to get, so, and you'll see why.
Let's have a look at some history. This is from 1990 until 2007. It's from 1990 until today, because all this stuff gets updated every day. Lots of curving, bottom left, top right. This is one is IANA to the RIR. The blue curve at the top is what IANA was doing. That first steep rise, 1990 to '94. We were giving away class Bs. Right. If you were alive and you know how to send mail to the Internet, they would get a class B. If you were alive and you knew how to send [] may you'd probably get a class A, but nevertheless, that was what was going on. Oh, my God, if we had kept on going, we would have got to the top of the chart by 1997. No sweat. So we did two things:
We did the temporary stopgap. That was called CIDR, and we did IPv6. Be aware of over achievement. All that CIDR, every last little bit. And you'll notice that even the Internet boom wasn't really a boom in address space. All that 2000, 1998, and so on, there is this tiny bit here, that's the Internet boom. As the [Telco] business always said, it's the trailing 50 percent of the market that matters; it's not the leads. It's actually filling in the rest. [Stop] in fact, the bulk of the work of the Internet is actually happened in the last three years. All that have stuff we heard of, deployment is now. Fascinating: Now, just when it's running out. So down the bottom there are the RIR curves.
Let's look at it a different way. No train crash, there is a new one every slide by the way. That one's a ripper. This one is actually what the RIR has handed out. The blue line is the total. It's a lot /SPHAO*LT err. The /STATS come in everybody single day. This is all public data by the way. So again you see the early stuff an then CIDR, relatively smooth graphs. These things along here, IANA used to allocate things and if you look in the registry you'll find these funny slash 8s, the old class 8s. So that's /TPHAPB /AEUFPLT then the RIR system kicks in in the mid nineties way down there. We have actually handed out very little space. (IANA) but we have fully filled most of the Internet. We have actually been astonishingly efficient since we came in interestingly enough. Our policies have worked. So we'll go down into those in not much detail.
The big three curves slightly above the other two, ARIN, RIPE, /AP nick. The bottom ones, lock nick and /AP nick, because they started more lately. Right now fascinating. ARIN is slowing down. Over the last few months ARIN has given out about 17 to 20 percent of the address space. RIPE too is showing admirable restraint, only about 25 percent. Where those people over in /AP nick, 50 percent of all the addresses in the last view /TPOPBTS from come through /AP nick. Interesting enough where the demand is being express it had on this planet right now. So the systems are moving slightly.
In actual fact in looking at pre /TK*UBGSs, I actually try and get, if you will, very, very consistent data and the best I can find is alley snap /SHOETS of what's being advertised. What I am trying to understand is can I get predict I have models that are based on relatively large amounts of data extending over a period of time which also show me friends in history. I am working on the assumption that address space that gets handed out /REUTS the routing table. I am working / PB the assumption that our policies work and no other bits and piece was data I picked up, this is indeed a valid assumption.
So here Yours sincerely routing table, not in number of entries, but in amount of address space advertised. And down the bottom, the /TKPWRAOEUPB line, is the amount of address space not advertised. (Green) so as you see, you know, it's the usual ex /POEPBG he thinks curve, down on the bottom of the left... some lig interesting variationings, we'll get into those maybe, some larger ones, those little square jumps, those are lash 8s. And they are actually /SPAPLers, what they do is they advertise 80 slash 8. In other words they cover the whole space that's allocated further down and then just occupy a few vacant slash 32s to do spasm. Down the bottom of course we also see the unadvertized space.
What we are trying to do it is do a predict I have model and what he have done is least swear best fits, one across the further order differential use ago polley /TPHOEPL /KWREL order 2. That's right. If anyone remembers maths there were some sense there. Not enough time, not enough space but you know apply your own model if you didn't like that one. This one seems to offer the best fit and what I have applied that model to is the advertised addresses. That red line is today. So if the left hand side looks like the right hand side, I have kind of got it right. If it doesn't, I have got it wrong. So, what's going on there is that the advertised addresses, as far as I can see, is going to increase subtly and slowly. The unadvertized address pool will continue to increase but at a much, much slower rate, ex observing the last few years as well and that gives us a picture of total address de/PHAPBLD. So we actually now understand how much addresses we are going to need to fuel this system.
Valid assumption: Well, you know, the last seven years have been amazingly consistent. Why? I mean why do people in very, very large numbers awful a sudden behave union formally. There is no massive variation from month to month. As far as I can see, the real issue is deploying the Internet costs. And money in large amounts moves slowly but inevitably. And what you are seeing is the cumulative effects of investment decisions by players day after day, year after year. And that cumulative effect is actually quite predictable. It has its own momentum. And as far as I can see, it's completely ignoring the fact that that right hand side line is real. /WAOPS. That's not a bat crash either. I think that's /KHA /KAG owe by the way. Totally off the rails.
So how quickly are regretting rid of address /STPH S there are only 44 slash 8s left. I am looking very, very hard at the RIRs and I have totalled all five. This is interests because in actual fact, you guys, the industry, go to sleep and you go to sleep about now. Maybe everyone who ever ever wants address space comes to these rooms. Because round October of the last few years, we have only been given out at a rate of six slash 8s a year at that point but when you come to back to the office, and in fact you really only start working in February, and in February 2005 you decided you'd get through an equivalent rate of 14 slash 8s. Now if you normal eyes that and do the least square bets fits over this differential, you will see we are get to go a run rate of about 12 slash 89s a year but increasing, so you have got about three years left. Again if we apply this, I can actually model the RIRs individual actions, because right now informally between the RIRs, we have all agreed, that when we /TKPWO to IANA for more we are going to get two slash 8s at this point. So I am modelling that. Down there at the bottom I am actually modelling each RIR down to its low point. It gets another two slash 8s and so on. That then gives me the demand picture from IANA. Because IANA has only got 44. Now assuming you guys don't do this who gets the last slash 8 rule business, as we run it down as it sits it's going to happen like that. Every time an RIR comes, it gets two and down there at the bottom is when it gets to zero /WHARBGS time does it get to zero? Well interestingly, this is just maths. It's not me. You know. It's least squares best fit to a first order differential use ago poll /TPHOEPL /KWREL order 2, remember? Of course you remember. That gives you this date. 22 nd May 2010. That will vary, that will vary. And my suspicion is about this point in time it's going to come back in again because you can never model a last chance rush. I can't do, you know,what happens when all of you currently decide oh I better go home and get some more address /TPR S RIPE NCC. I can't do that. Can't model T so at the moment as long as you are all well behaved, as long as there is no panic, as long as you sit there in the address policy meeting and say no, no, not going to change a thing. As long as we actually don't change any of the underlying demand die /TPHOPL /EUBGS. As long as no one steps in to help you, because you need help. That's not a bad crash either by the way. I think it's in I say stand bull.
As long as there is no rationing. As long as you don't decide all of a sudden you are going to do the soft landing or the medium landing or the can't land at all and as long as there is no /HAORDing or with /HRAORDing going on here. As long as you guys behave like no other group has ever behaved in the past (/HAORDing) then that prediction will hold.
So, you know, let's now talk about the after life because quite frankly, I won't have retired by then. And looking at you, you have got a lot more hair than me. You are in deep trouble. So, perhaps we ought to look at this and figure out what scenarios are there.
Well, these are not either ors. You may have to do all of them. And that's going to be not a lot of fun. Because quite frankly /TP FRBGS there ever was the word addiction, then v4 and addiction go hand in hand. Remember, CIDR was short term. You were meant to have done v6. You haven't. Somehow the signals haven't got through. And if you haven't got through now, they are never going to get through as far as I can see. Persistence in v4 technology. How much money has this planet invested in v4 in dollar terms? We are talking about the most valuable industry we have ever invented, multi /TREULS of dollars investment in public global communications. How much that have investment is in v4? A massive amount of it. There is been a lot of /ROUTers you have bought, /SEUS co shareholders are happy people. You have made them that way. You are probably going to find some kind of market somewhere and I use it in the broadest possible sense. The market is really you know, I have something you want, what are we going to do? Barter, exchange, trade, whatever. And that sort of markets in my mind, there is some kind of different distribution mechanism, because if you are addicted and if the current supply mechanism no longer can supply, what are are you going to do? Addicts will always find an answer. And you are all addicts. So, what else are you going to do at the routing system is a common victim. That's just all over. Plan for bigger routers F you are going buying equipment that's going to last for five years in the field, let's just understand that five years from 2007 extends beyond 2010, doesn't it? Right. Buy bigger ones. You know, big clue. And you might want to consider v6 but let's have some fun with that.
So NATs today, I like that one too he is on a train NATs were an unintended amazing success story. ISPs don't pay for NATs. You do. So the whole current cost of what one would say is address scarcity even today, these policies /R PBLT tough. You can't get what you want. You actually have to justify. You actually have to go through a fair deal of work if you are an ISP to deploy /AEU /TKESs and typically for retail residential customers you hand them one address and say have fun. What fun do they go out and go out and buy leases and do what, NAT? Do you pay for it as an ISP? No whose problem is it if not yours, you have external eyesed that cost. Fantastic. Not your problem. An application folks from gone oh deer, okay, let's make the application better. Talked about it this morning. Understand what's going on F you want an application to deplough into today's Internet and it doesn't work over NATs, what planet are you on? Because it's not this one. Okay. /*FP
So what's going on now is that right now addresses are a market. I am a residential broad band customer and I want a permanent IP address, how much do I pay? The going rate in the market today is at least 10 dollars US per month per address. Oh addresses aren't a market. They are. They are being sold. 10 dollars retail. What's the wholesale of price of something that sells for 10 dollars retail. I am talking about the Telco industry mind you. We are talking about larger margins than normal. So you know, 50 cents an address per month. So how much is that a year? What's the capital price of one single address? Start to do some sums here because when you start to understand that, addresses have considerable value out there right there in the market that you operate in. Because you you are already selling them to consumers.
One of these options are, push the train offer the edge one of the option /R S just add more NATs because there is a demand out there nor increasing NAT intensity. At the moment you have only just made the customers do T you can do it yourself. Of course you can. And you will. And you are going to start shifting this into infrastructure. Rather than just all the customers doing it, you might find it in the POP, the first point where you still haven't got massive/WAEUFL
Wavelengths of traffic you have only got smaller amounts you might find you can deploy NATs to do a little bit nor funki stuff and you are actually going to see, as far as I can see that you are going to look at multi level NAT deployment out there near the edge as well as right on the customer's premises. For applications this becomes a little bit weird because the one thing the IETF did in its /PWREUL /KWREPBS was decide that notice should never be standardised and the one things NATs are is very, very tricky. If you, as a code err, were told to build a NAT, then you start scratching your head about U D P because there is no rules and because you are a code err you are obviously creative, you go and invent your own solution but the guy in the cubicle has done it differently and the next one has done it differently. It's not that NATs are evil. There is no such thing as morals in coding. It's that they are variable and that's what makes it hard for applications, guys. If all the NATs behaved according to one standard we wouldn't be having this debate about the morality of NATs and it seems weird to me that the IETF that dropped the mull so /STUPBGly from such a great height says that NATs are evil when it was their fault, they /TUFD it up. If they had standardise it had, it would be a /TPHRAOEL different argument. Just add more NATs so matter what. We find...... standard behaviour NATs the industry is really, really going to get them. And I think it's really, really going to happen.
So, you know, NAT futures are there, they just represent more of the same and the /WUPBG thing this industry is is it's not inventive. It is not creative. A bunch of people like you and me doing what they did yesterday tomorrow. So this /KAOEUPBLD of large and relatively dumb industry is sit /THR G going oh well, more of the same, I can do that. Don't ask them to be creative. Because people in large numbers like the Telco industry behave very, very weirdly. And if you think they are going to graph /TAEUT towards v6 /WHOEPB you give them a creative exercise I don't think you have really studied the had I tree of the Telco system. They will do anything. Even O SI.
How far can that scale? We have no clue. We have never tried. We are going to find out though and perhaps we are going to investigate what the real critical resource /R S in NATs. How much can we make NAT binding capacity and state magnificent nets work in we haven't really stress it had. What about NAT packet threw put? You are going to find out. What about address pool sizes and application complexity? We are going to explore all this space. It's obvious.
So you know /TPHALGTS on steroids, of course. Because when you start thinking about this v6 transition and protocol translating NATs, you are going to be stressing the model anyway and we are going to think about how to die in the DNS so you don't just /TPUTS around the destination address, that's too easy, I can muck around with the source address too, I can rewrite large /SHRABS of this packet. Then there is a case why just wait for the packer? Why not make NATs eminently predictable. Why not do NATs signalling protocols? Because if I play a game I want a concern signed of flats behaviour. In I want a voice over IP I want a different kind of behaviour. Why don't I scream at the NAT. Why don't he scream at the /TPHALT. It's going to /HAFPLT at some point we start to create /PHRAEBGS tee because we are good at that. So much complexity that we come off the rails. At some paint you say v6 really was a lot of ease /KRER. How far do you need to go in order to get to that point? Again, maybe we are going to find out.
Randy bush said earlier this week it really was a tragic mistake but IPv6 isn't compassable with IPv4 on the wire. So, this is a real problem. Because when we are trying to do v6 and v4 when we recognise this massive /AEU /TKR*EUBGS to v4, v6 is not just a light bulb you turn on and destroy the other thing. We are going to run both. Dual stack is the only answer they have come up with. Somewhere is dual stack, I put it on all the host. Everybody single host. That's a bit weird because I am about to run out of v4 addresses. Oh, dam. So, maybe I have to do proxies and dot translation elsewhere. Maybe I already don't want to upgrade my network with v6 when my customers are giving me a margin of 50 cents a decade. Maybe I need to do something else out there. I think we are going to see not just dual stackings in the host but certainly /PHORBGS of the trance lakes proxyings
/TPHEPB re try to start to blend the stuff on the rails. (Then) so the first idea is that one rail is on top of the other. The theology phase one is you do the v6 at the either one and do the tunnelling. We have seen this before. At some point we manage to get the /TPA*EUTD and we all start convert to go v6 and we get mixed transitions, mixed slingsings and then at /SOL point you enter into the are the world is v6 and you shut down v4 and we have all had a good time, right.
What are those assumptions? They are pretty breathe taking because most of them are wrong. The major assumption was that all of this was meant to happen long before today. That we could actually do this while there was still v4 address to say actually fuel the network. Because dual stack really did mean dual stack. I needed both address families. The other one is that this was meant to happen before we got successful. Because it was only going to meant to take a small number of years. And these days trying to get 4 /PWEURBGS AS numbers in the favourite router of your choice which is a code change of only a few hundred lines takes five years. If anyone ever uses the word Internet and agile, they have been asleep for the last two deck /AEUTSDZ. We are a massive business. Change is what we despise. We have got a problem.
The other thing is that this transition was meant to be driven by the market. You guys were all meant to figure this was in your own best interests and your local decision would say become global. So all of this was meant to happen before today. /WAOPS, because dual stack isn't either/or, it isn't magically going from one /OT other. Dual stack is "And." The next ten years of Internet growth needs v4 addresses. Dam. /TKA*PL.
What was meant to happen was that as the Internet /TKPWRU and we saw /SH shortfall in v4, we were meant to be deploying v6. So that when we got close to the bottom, it was all over. /AOPS. That's not a bad /AOPS we were meant to have done this before the two trains collided. So now, in reality, there are 923, maybe 924 entries in the IPv6 routing table. There is probably for neck nick deployed, certainly more SN A deployed, even probably more apple talk. So why v6? It enjoys no better current deployment than almost any other protocol that isn't deployed either. Quite frankly, that's a zero line. So, get the difference between theology and reality. We are really in a very bad place.
Whether you like markets, /TKES pies them, no matter how you /P RPBLey feel about T this industry, the industry you work for needs solutions, because they are in a very, very bad corner, and the train has indeed crashed. Because you are going to need v4 addresses long after the RIRs have run out of their unallocated pool. /WAOPS. And the other thing we did was that we had policies that enforced /PWAEUFRS by simply saying if you want addresses, these are the conditions. That control disappears with the last address you hand out. Because now, after that, you have no control. If you are a bad boy you won't get more addresses. Stuff you you, you haven't got any more. Where is the control mechanism in that? Once the pool in the sky once dry, the world changes. Your world changes. And trying to understand the nature of that change and how many desires, how many shoulds get replaced request must? To what extent do your desires get dictated by imperatives? These implication /R S deeply shock to go a community that used to think it it had control. You are now victims of economics. Because quite frankly, there is a few other bits an pieces out there.
It's not the /AEU /SPWRES /SPAEUTS that's been fully consumed, it's just the unallocated pool in the sky and remember those 46 slash 8s, there is what I think called liquidity. You know, there is a certain ah supply out /THR FPLT and then we start actually counting addresses. Folk have done various things. There is one a couple of weeks ago recollect pinging in the network, lots of cul ears an curves but underneath it was a very, very strange store /AOEFPL when you actually start to count hosts, you don't get to 4 billion, you don't get to 1 billion, you only get to a few hundred millions. In actual fact we say 80 percent of the all the address space should be used, it /AEU /PAOERGS we have been been a bit naughty and telling lies, because in actual fact you have down around 5 to 20 percent even if you do the most liberal interpretation of pings an everything else. What we see as visible in the network recollect outside the NATs, addresses that are used in switching, we certainly don't get anywhere near the numbers of blocks being advertised. So, you know, second life could exist.
It would be a bit /TKAOEFient, the radiation would have changed the /SKWRAO*EPB pool, it's not life as you knew T absolutely not but you can certainly understand there is liquidity out there and it's simply a case of we know what you are, it's just the price we are arguing about.
Now I have seen a lot of /TPAOEBG with a lot of ideas about this and you are probably going to hear a bit more this week. Here is the list of some of them.
Encourage NAT deployment. I say the real thing is encourage NAT standardization. Creative programmeers should be shot because the one thing you really need, right, are NATs that work predictably. Once you have that the whole problem of application complexity doesn't quite go away completely. But by God it makes it a lot easier F I know the way the NAT behaves, then all this /STUPBG, turn ice and so on starts to disappear.
There have been proposals to free off 240/4. You haired about that yesterday as well and I am indeed one of the conspirators in one of these proposals and yeah we could probably buy a bit more time. I am not sure what we'd use it for but it's out there. There are various policies to ration the remaining v4 space. Rationing is kind of auto weird. It's always the principle of maximising unhappiness. I didn't get what I wanted. No you didn't get what you wanted because there are other people in the cue. We have to give them some. (Queue) but those people aren't going to get what they want either. I want add whole slash 8, you are giving me a slash 24. I am not happy either. Under rationing who is happy? No one.
Maybe we should figure what to do with the last slash 8? I have an answer: Give it to me. I am glad people agree. What I do I promise is to put it onto the market straight away to do the necessary liquidity because we are discussing much the same stuff. Let's encourage and accelerate the IPv6 transition process. Be careful. Be very careful of what you ask for. Because you might get it. We'll come back to that.
But anyway let's talk about making IPv4 last a bit longer. There is a number of real questions out there because how long? What's the cumulative demand? What's the level of fairness? At what cost and for whom? To what end? But, remember CIDR, remember how good you were. Remember that you bought. You bought 13 years when you were only meant to buy five. You over achieved. You succeeded beyond anyone's dreams. What if you do it again? What have you really are good? What have you really do manage to make it work? Would you be a victim of the law of unintended consequences yet again? If you can make it work and I am paying the bills for an ISP, I am not going to run v6. I am not going to pay the price because if you make this work, the entire IPv6 as a solution doesn't work any more.
So, be careful of over achieving.
So, you know, inside all of this, and it's a lot of fun, and there are some very good crashes, inside all of this don't forget there are a lot of people depending on you to be able to pull this off. Because you don't really have ones to worry about this. You know, what's v6? What's v4 (what the hell are you talking about? Where is my browse /STPHER I want to get to Google. You need to do so. You need to preserve the functionality of applications. You need to put all these balances up in the air and keep them up there. It's not going to be easy because you have got to make routing work as well (balls) and the world isn't going to stop its growth just for you. You have unLaoised a massive multi billion doll other investment programme and it's not timed to stop in 2010. The network will continue to roll somehow and you have got to make it work somehow.
So, you know /WHARBGS could be useful? Information always helps. Right. We actually need clear and coherent information about you know what the situation is and the choices and I hope I have given you some today.
You also need to understand the implicationings of some of these options, right. There is one down there at the bottom. We'll get to that.
We're a deregulated industry. Saying you should move one room to the left won't work if I simply say it. The only way you are going to move to that room on the left or the right or your left and my right is if I somehow induce you, because the deregulated industry only works from two stipple lie: Fear and greed. And unless one or both operate, you ain't moving. So either I have got to do v6 because my competitor is doing, oh my God, or wow, think of the bulk etc of money. And if neither of these things are happening, the city won't listen. So our strengths and weaknesses are yes we are adapt I have, /KWRER we are change, yes we are fiercely /KPHET I have but we only respond to two very, very crude stipple lie and if you don't provide either of those, we are not moving.
Understand too, that governments are out there. You are a public telecommunications network for the globe and you think governments don't care. That's their role, public telecommunications for the globe. How many more key words do you want?
And there are some pragmatic choices out there, but just understand one thing: That most of the time when hard decision precise taken by this industry, it wasn't the process of market economics. It was the process of direction by regulation. The digital cellular network wasn't digital because markets decided. It was a case that the industry was told and consumers were told the hand sets you are are now crap. On this day we are going digital. See you later. The television industry is going digital. Why? Because they want to? No, because they are being told to. Be careful of what you ask for, because others may interpret it as a signal to direct.
You might get it.
So, as far as I can see there is some amazing implicationings of this particular egg. Recognise that v6 today right now today is a business failure. Whatever its technical aspects, for business it's a failure.
And resolution of that failure is going to be painful. It's going to be very, very tough. It's not going to be seamless, it's not going to be costless and perhaps you are not going to respond as a market, as a deregulated industry. Because as far as I can see, if the signal from the, if you will, the industry S oh my God, we really, really need v6 but we don't think we can get there, there are people out there whose job Yours sincerely to listen to such signals and help you. And direct you and instruct you.
So, you know, we have been here before. We are very good at this. We even invented if you will a standard set of phases industries go through when they are con/TROPBTed with crises. Same thing as we as individuals go through. Massive problem, sky is falling, no it's not. No, the sky really is falling. Oh my God is it really falling? Anger: Bloody hell who started making it fall. Blame shifting: It's your fault.
Then you get into bargaining: Make it four, fake it fall over there please, I'ding good, v6, make it fall over there. At some point you have got to recognise that yes the /SKAOEUL really is going to fall and it's going to fall on you as much as me. It's the sky. It's falling. Accept it. Get over it. And then do revisionism. I knew it all along I was just playing with you. Where are you? You used to be inney /STKWREUPT on a river called the /TPHAO*EUL, but no, now you are starting to get the answers and get the clue, now you are in panic, good luck. Thank you.
(Applause) (Egypt) /ROP /ROP thank you very much Geoff. Are there questions? It's your union mates.
Question: Hello. I am his union mate and actually I don't have a question. I just wanted to make a suggestion and that's that the last shade is given to the RIR sheaf scientists union pension fund. (Chief)
Answer: Annexe /HREPBLT suggestion. I second it.
Question: So you said the IT E F isn't going to standardise NATs which I have also seen, heard and watched the /TPHARBing of teeth. So that means they're going to get standardised someplace else. Where?
Answer: That you truly is an amazingly interesting question, because the body that does, is about to be right in the core of tomorrow's new Internet. So, it's not really an oh my God who might do it? It's who is going to race to do it? And I think you and I both know, rand and probably everyone in this room, who is going to be in that race, because there is a /TKPWHRAOPLG understanding of precisely what we are playing with here in terms of bodies, standard iceation value to industry. This is a change from where we were today. This second life is writing a lot of new rules and the level of adopt /TAEUGS and agility that's being required is going to stress all of us, including the IETF. But they are all being warned. I think the information is clear and apparent. Now it's a case of who is listening to those signals and willing to react. So /REPB, I have no suggestions other than everyone else and one of them will surely do it.
Question: Dave Wilson, H E A net. Listening to the last three talks, if I were to try and leave my prejudices at the door, I would take a message from this which is: If we attempt to create in our own image an after market, post 2010, then we invite some sort of regulation from the outside. And if we attempt to not create that market and try to avoid its creation, then we end up in a situation where we don't know what's going to happen and we invite some sort of regulation from the outside. Can we actually avoid it? Or indeed, should we be thinking about how we want to engage with that?
Answer: That's precisely the take away that I am trying to offer and maybe I should come out and be perfectly overt with this. You can't play with global telecommunications lightly. And you can't create these kinds of crises and they are cry session and think that they will magically solve itself without the involvement of very many, many people and you can't have the arrogance to think that you can devise a solution here in this room with only yourselves and believe it's the appropriate one. They are actual' a whole bunch of other parties, other knowledge, other interests that have a stake here. We need to be talking to economists. We need to be talking to public policy people. We need to understand that everyone has just a part of a picture. But in order to make this work for users, actually have a network at the other end rather than the old U, net, bit net male chaos we enjoyed in the 1980s, to have even a built of what we enjoyed today, we are going to have to create partnerships and alliance /W S folk we have never known yet. So I'd encourage all of us to start talking to you know other people. To actually understand where more knowledge and more perspective come from, because to think about all of this as just NATs or v6 misses the real issue about industry, regulation, economics, public policy. Interests around son /SAOUPLers and indeed larger global /SERPBS and they are there. They are here (consumers) and we don't engage in talking to them, they are going to solve it anyway. They really will solve it anyway. So yes, Dave, that's the message.
Question: The first thing that comes to my mind is that probably after this presentation, another session will be added to the AG F, a session about trains. The question is, as usual your presentation was very good, very interesting. I would like to call you on two points. One of them is about the possibility of increasing the usage of NAT. The NAT has been used for a long time, very intensively in some parts of the world more than in others. So, the number of IP addresses that have been allocated in some regions for Internet users, and please I am not saying population, I am saying Internet users, is much smaller than in some parts of the world than in others. So if the alternative is if the option becomes to make a more intensive usage of NAT, solve the problem that the people will face will be different in some parts of the world than in others (so) because the margin that they have in some places for still increased the optimisation of the usage of the IP L address that is they have is much reduced than in other places. So this is something that we have to include in analysis of this option.
The other thing is that you mentioned when you listed the things that are being discussed or proposed, to face the IP population. I think the ration /HRAOEUGS, the accession to IPv4 addresses in some way, I am afraid you see this only as a way to postpone the end of IPv4, and that it would be also negotiated as a way of discouraging the usage of /AOUFR and stimulating the companies to move to IPv6. Thank you. (IPv4)
Answer: The issue about dual stack is it's not an either/or. That unfortunately dual stack really does mean dual stack and for as long as we need to do that we have actually got to /THO do v4 jumping through the hoop to make the v4 part of dual stack work. And we don't have any magic out there in technology that gets us past that particularly unfortunate truth. So, as much as it would be so good to say it's backwards compatible on the wire, you can deploy v6 only and get away with it. You can talk to the rest of the world. It's okay. We can't do that in the technology. You are forced to somehow do some magic that let's you talk to v4 and whether it's a NAT P T at the front A protocol translation or not, by the time the packet comes out, the packet has to be of a v4 headed with v4 addresses and those v4 addresses must be be public to hit the core. So there is no magic here unfortunately. The situation is indeed quite bad. And the current distribution of addresses is relatively skewed, and a market ex as err baits social issues, as we well understand already, as much as is might solve some problems. And this is /TKPWR*EUS for the mill for typical public policy problems of a global scope. So it's not as if the world has never encountered these before and I think part of the fear indeed in here is the world has encountered this many, many times and has well reherringsed solutions. It's just that we are not used to them inside the Internet community. But our broader industry plays this game for breakfast.
Question: I see your point. I agree with you. Anyway I think that it is something that we have to include in analysis and also it's something that we have to keep in mind in the time of evaluating all the package of possible solutions of measures. Thank you.
Question: /TPHAO*EUL Murphy. Google. Thank you for your selection of illustrationings of systems failures.
Answer: I have to give due credit to George Michael son a dedicated steam train nut for these photos.
Question: It's clear that many of the important people aren't in the room and it's also clear that we are sadly lacking a vision. I guess, thinking about the last thing you were saying in terms of we need to widen how we do consultation. I am just wondering recollect given the time frames we are looking at, consultation is a time expensive process. What guarantee do we have that within the exhaustion time frame we have consulted with somebody who has a sufficiently useful idea?
Answer: As time gets shorter, you will get more desperate. Would I start early /KWRORD to be able to critically evaluate what you are hearing, because if you leave it for another two years, anything that floats is going to be used, and you may not you know you may not like it. So you know, even now, the earlier we understand and start seeking outward recollect the more we are able to critically evaluate as an industry and as a policy body here, what are sane choices because if you do nothing for two years you are going to get awfully desperate, you are going to get awfully panicey and the solutions are probably going to be awfully random.
Question question: And awfully ugly.
Question: Thank you. That was very entertaining as well as scarey. I do talk to public policy people. That's my bag. And when occasionally they talk to me about Sykes it goes along the lines about we have heard about this (IPv6) and somebody has set /AUP project to encourage all you people in the industry to move over. What can we do to encourage you to move along? And what do I have to say in response to that because you know, I don't even I am a membership organisation, I don't even have a company, a single company that can move over. And the answer S there is not really much you can do to persuade us because we are not going to move over as long as we can keep getting IPv4 address space. Now, what interpretation do I put on this own response that I have got? Your statement that IPv6 is a business failure doesn't wash, because your own, by your own analysis, it was never going to be taken up now because there was no reason to move to it. It's only when we come to the train crash that it becomes tested. It hasn't been tested yet. So that then we have to ask: So what does failure actually look like? It doesn't look like your little graph showing that nobody has deployed it because that was never going to happen. What failure actually looks like is, if we run out of address spaces and decide well we have got to move to v6 in a hurry and find that we can't. That's when people start getting really panicey. And so I think where we need to be turning some of our attention in the analysis is not to oh God they are going to run out soon, but, instead, how quickly can we deploy this if we need to do so in a hurry and are the points that are going to feel the pressure first, the access networks, going to be unable to deploy it effectively because of things further up the chain? That's where some really serious work needs to be done, in my suggestion.
Answer: Thank you. Let me respond by pointing out that this isn't the only pool of numbers that was subject to exhaustion. A small pool, finite space that is being consumed. The other one you might recall, some of you, AS numbers, autonomous systems numbers which were only 16 bits long, 65 thousand and way way back in the year 2000 we understood that we were going to actually run out. At the time, because it was the Internet boom T looked like we were going to run out in about 2005, /TWEUBGS. We did the work. We did the projections and then understood we actually needed some technology and some very, very bright people, who I think looked at v6, put forward a proposal around putting larger AS numbers into BGP backwards compatible, backwards compatible. So that I could deploy it and none of you would have to worry. This looked good, but then the Internet bust came along and all of a sudden the numbers blue out again. Then we did the analysis and found it was 2010. 4th November as I recall. And it's still the 4th November, 2010. AS numbers are very /ST D /STED owe. We started three years ago, three years for a backwards compatible technology and fist we have put through the presentation to vendors. What did the vendor say? Look, it's fascinating, I am really amused, this is very, very great fun but quite frankly until a customer, and not just one, but a huge army of folk come to us demanding this, we are not going to build it. Because as vendors we respond to money and the money isn't there, we are going to do nothing. And you sort of sit there and scratch your head and say that's not a response I want to hear. So then we come to the community of the addressing folk and we say look, how can we make this happen? How can we actually encourage signals? So we create policy and as you might recall the policy that was we'll hand out the 4 bite addresses, now in fact we started this year and if you want some you can get some but if you don't do anything recollect you'll get a 2 bite number. How many people have got them? 35, 40, very few. But all of a sudden we have having conversation /W S vendors saying are you serious? Why are we having this conversation? Because, the other part of the policy was that on the 1 st January, 2009, just a little under two years before we know we are going to below T we are going to (blow) flip the bit and if you come to RIPE or /AP neck and say I want an AS number you are going to get a really big number and your router better cope because at that stage you are going to need new software. Only now, four years later, have they finally at the project manager level said ah, your serious and the answer is, yes. So the next question is how long have we got? Five years? What? So we are now talking about a change to BGP. There is not many routers out there with BGP. It's not in your every single host system. Your question is if we really need to flip from v4 to v6, how long is it going to take? I am not going to live that long. Seriously I am not going to live that long. This is not the network we group up /W FPLT this is entirely different. This is massive. And the logistical issues with sub chains and everything else. My God none of us are able in software project with three people are able to get the estimate right. Yes? Have you ever done a /SHOFT wear project on time in budget and right on prediction? None of you are nodding. Neither am I. This stuff takes time pause you are dealing with an awfully large complex set of players and systems so, no, if you wait until the end, oh my God.
Question: Thanks. Could you slide back to slide 34 Tom vest, consultant to RIPE NCC
That one. Could I make a pitch for everybody thinking about this to adding a fifth one openness to new entrance. If you look around there is a lot of people in this room just from one region. Probably more in this room than there are ITU planet pods, certainly more than that are ETNO
Answer: What year is the next IT N O planet potential tree? 2010?
Question. In any case the point is we have manage today pawn a system somewhere between 20 thousand and 200,000 independent entities can demand a top level spot in the control plain. Compare that to the number that are you know, candidates for being planet posit the I /TURBGS, and again, that was that was the control system and the industry structure that we had before we had this logical control system which you are addressing so important. I mean everybody in this room, there is a lot of people this this room but number sense we are all /K BTS. So we actually think about the rationale for the Hostmaster process, the needs based allocation and the idea was that you know, you get what you need today because anything that you take in excess is in effect depriving the future of an opportunity to do something interesting. And I think that's maybe the most important thing we need to preserve at a very high level with whatever comes after this. And I would commend you to add a fifth bullet to that.
Answer: I /STAUPBLD the point and thank you. Hopefully this is the last time I ever have to give this presentation so /AP nick and RIPE are privileged people.
You have all heard the message and you will go and do it now.
ROB: Okay. Last question, Daniel.
Question: It's Daniel again and I am organised. Just to answer row you'll's question about who is in charge of the trains? Actually, when these things happened more and more, trains got heavily regulated. It didn't prevent all the train wrecks and of course the regulators were blamed sometimes.
Answeren. . I think they deregulated in the UK, what happened then? Some got LoST and some went off the rails.
Thank you very much.
(Applause)
ROB: I think we are done for today, at least the serious part. You are all welcome at a social starting at seven o'clock, kindly offered by one of our participants, and I hope to see you all tomorrow morning at nine o'clock went we continue. Have a nice evening.
THE CONFERENCE THEN ENDED FOR THE DAY.
|
|
|
 |
| RELATED TOPICS |
| RESOURCE LINKS |
|
|
|
|
| |
|
you
are here: RIPE -> RIPE
Meetings -> RIPE 55 
