THE PLENARY RIR/NRO REPORTS SESSION COMMENCED AS FOLLOWS:
CHAIR: Hello, if we could just take our seats, we'd like to start. Thank you. Hello everyone, welcome. My name is Paul Rendek from the RIPE NCC and I am chairing this session. So first up, without any delay, I will introduce my colleague Daniel Karrenberg, who will be talking about the latest project we have got happening at the RIPE NCC called RIPE Atlas. Daniel.
DANIEL KARRENBERG: Thank you, Paul. Welcome. I have 30 minutes and 35 slides so this is going to be fast. As you know we are doing a lot of measurements and measurement infrastructures at the RIPE NCC and one of the ones that we are currently using and offering is RIPE TTM test traffic measurements which has been around for a long time. You can see a map of all the test?boxes that are out there around Europe and the greater world. If we go and compare this to a picture of where actually the Internet is and this is courtesy of NASA, thank you very much, distribution of light in Europe and surrounding areas, and my conjecture is wherever there is manmade light, there is probably man?made Internet. Then, if we compare the two, there is a difference. Thank you for turning the light down. You can turn it up again so people can continue reading their e?mail.
So, we are thinking about, if we want to make maps of the Internet and if we want to make, look at how it's functioning in the broadest sense, then the ? are probably not going to give us a very detailed map, so what we did next is we a did a couple of simulations to get some intuition about what was needed, so what we did is took all the address space that was distributed by the RIPE NCC in the past and we took a statistical sample, a random sample of it and if we ?? and then we used geolocation to see where these things are. And that work was done by Emile, and he did a good job, I will show you a movie in a second. What you see then is if you take a random sample of 1,000 address blocks, then it looks like this, and this is already sort of a little bit more approaching the light picture that we saw but it's by no means there and what we did here is we put a probably in there, all the data basically the colouring between sort of green and red, is random, but in one country. And with 1,000 probes you can't really see which country it is. If you go to 5,000 probes there is a little bit better picture already, I am not sure whether you can see it, but if we take a few more probes, like 10,000, it becomes quite obvious that it's somewhere in the Baltic area, and this was sort of inspired by a historic event. So, what we did here is we just took all the address space that the RIPE NCC distributed, took 10,000 of those blocks or prefixes, if you want, randomly, and actually, this already approaches the light picture that we saw quite well.
If we take 20,000 it becomes very obvious, right, where we are, and if we take 50,000, well, it doesn't add so much. And now I hear you all thinking, oh, he does something on geography and the Internet doesn't follow that it's a toplogical thing. So the next thing I am going to show is you a movie that is actually another simulation, it uses a number of probes and I am not going to tell you how many just yet, and what we did there is we inserted a problem into a single AS and let me run this movie. And there you see it's all random and they go like on and off and red and green and all that stuff and around here there is a problem in a single AS, and that gives us sort of the intuition, what we could could do if we had so thousand of these probes and my aplogies to Rudiger for picking that AS, it just works so well.
So here is our intuition, so we say, OK, we probably don't need 50,000 although that is about the number of ASs; but we need, you know, 10,000 probes, coming from 100 that is two orders of magnitude and that means we have to really think very hard and once we have to think very hard and here is the slide of the movie, we go for inspiration and we go outside and go for a walk and many of you will know this, the site that is from the Krasnapolski, the usual RIPE meeting venue, it's the royal palace in Amsterdam and of course we don't go for inspiration for for the Royal Family because they don't really live there. And then we go for some comic relief and if we see the palace, the hard work and the heavy lifting and the real unthankful work always happens in the back, this is the back. Here is this guy and we are talking about maps, he is holding up this ball and he is called Atlas and Atlas is also a word for collection of maps so that gave us some inspiration and actually you can see this place from a window at the NCC, and there is a webcam there. It doesn't come out on the projection very well but at the moment this back facade of the palace is in a scaffolding and you can't really see Atlas very well, you can maybe just see the orb that he holds up and very many people in Amsterdam were wondering what is happening behind the scaffolding. Thanks to Wilham's photographic abilities we can now reveal what is really happening. And if you look very carefully you can see that actually what I would call the legacy orb is inside the new logo, so, sorry, Paul, for revealing our new logo a little bit early, but it was on all the slides so what did I do? Right. Coming back from the sort of digression here, so we have to get 10K of those probes out there. How do we do this? And you want even maybe some more than 10K and if we look at it not only in the RIPE region but globally, certainly more than 10K. It's something that is intuitively something you could do if you had a lot of money, but it's much better, I think, to organise this as a sort of community effort, so may propositions here instead of building sort of small separate individual and private measurement infrastructures, is to there is a huge common infrastructure, oh, yes, socialist ideas, I hear you thinking, and what is in it for me? Well, actually, my proposition is to build this to serve some community goals, to get good maps, get realtime situation awareness of what happens in the Internet as a whole, take the big picture, in addition to the local picture that you have when you monitor your own network. That is all apple pie, I hear you thinking why should I contribute to this? We can also use this to serve our private goals. How about this: We give you, also, an opportunity to measure, to specify some of your own measurements that you want to do, to run your own monitoring that you always wanted to but didn't have quite have the energy or the means to actually build up a diversified measurement infrastructure. So what I am proposing here is we go and actually build something that serves both purposes. So the individual benefits would be it's less expensive than running your own, you have more vantage points than you would ever have because I don't think many of us would have the opportunity to build something that has the order of tens of thousands of vantage points and you have more data available and more diverse data available for your own measurements that you want to do and there is benefits for the community because we can get some unprecedented situational awareness from a wider distributed number of viewpoints and there will be a wealth of data as well, so there are many PhD theses to be written. How do we come to a plan? For an accurate map you need more probes. If you are thinking about making maps you need a lot of trigonometric points and the more you have the more accurate maps so it's the same thing here and deploying that many TTM boxes, which are sort of hosts that need sort of GPS antenna and all that stuff, is probably not going to cut it so we need smaller probes. We want something that is easily deployable and we came up with this little thing here. It is USB?powered but there is no data on the USB, it's just there to get the power because we didn't want to make something that has to have you know the 6 or 7 different mains plugs that are there around the world, you can everywhere you can get a USB power supply or find a USB socket, so it's USB?powered and the idea is that it's 24 by 365, you just plug it in and you forget about it because it takes very little power and it just runs there. So here are a couple of pictures of possible deployment scenarios. On the left, you have the office setting, you take a USB power supply, plug it into the network and off you go. On the top right, you have the coms centre environment and note there are actually in this picture taken by a non?techie, taken by one of our coms people, but just by pure coincidence there are 3 slots, USB slots in this picture. So that is what we came up with.
What can it do? Well, it's version 0 and I will tell you a little bit about about the genesis of the project a little later. That can do a couple of pings to fix targets on IPv4 and IPv6 and trace route to upstream hops so we have some comparison material and also you can see whether ?? if you install it at home, see whether your broadband connection with reach the second hub. In version 1, which we will ?? which will be coming really soon we want to be able to do ping and and DNS queries to variable targets. And then version 2, well it's up to you. What should one measure? And note that the up grades are automatic so it's not like you have to exchange the hardware or something like that, it will just update itself through the network.
Well, we cannot be everywhere without your help so this is really a plug to become a host of one of these things and donate a fraction of your bandwidth, just a small fraction, and at the moment it's not configureable, but we will make it configureable, we will let you say OK you go up to this many kilobits a second or something like that, and donate a little bit of electricity, about as much as a small USB device can draw. What you get in return: you get recognition so we make a list of all the people who helped and want to be listed, and you, at the moment you get the access to the fixed measurements from your probe. Well, it's nothing exciting yet because you could do those measurements yourself with a host but on the other hand, it does it 24/7 so you don't need to keep a computer running to do it. But in sort of the second quarter of next year, we want to go to give you measurement credits and what can you do with those? Well you can actually do measurements from any probe in the network. The reason why we do it with credits is we want to sort of ensure some fairness so that not one single person or group can monopolise the network but in the end so if you run a probe it will give you equivalent access to all the probes that are out there so you get more viewpoints than you could do yourself. That is the idea. So here is how the website could look like where you configure measurements and this is actually the map of just part of the RIPE service area with the blue markers showing people who have already registered to receive such a probe, and the green ones, not the one in Rome, other ones of the probe are up. We took delivery of the first batch three weeks ago and you will notice that a lot of green ones are around Amsterdam, in the Netherlands, that is because a lot of the RIPE NCC are actually helping us to do the initial testing, but there is one up here in the meeting at work as well. And if you have registered, you can actually pick one up later on.
There is a graph that shows the numbers, the blue line is the preregistrations, the green line on the bottom is actually the active probes that are up, and we have a target of reaching about 300 active probes for our initial feasibility test. That will be just ?? already useful for measurements, it will not be as nice as the 1,000, 5,000 idea, but it will show us it can be done and that we should move on.
So, what we expect to happen is that, today, there will be a sharp upturn because you will all register and actually if you look at the site it's you can see that the upturn is already happening so word of mouth and Twitter and all that stuff already is showing an effect here, and then we expect a really sharp in the green curve when you all get home and plug these things in.
Also, you know, to be sort of really commercial, I have got our finance department to liberate an iPad and the host with the highest up?time by December 31st will get that iPad, and so there will be around 300, so there is a chance of one in 300, actually the chance is better because RIPE NCC staff cannot participate in this little game, neither can board members, by the way. I am assured there are legal reasons for this. So if you get this we have to make some fair rules for this because, of course, there is a factor here in how fast you can get home and we don't want you all to leave the RIPE meeting right now to plug in your probes, so and we will discuss the modalities and the rules for this after the MAT Working Group over a beer.
So here is the alpha version of the page that the hosts, once you host and you register you have a page that shows you whether your /PROEP is up or not and things like that, and here is a part of it that shows you the results that actually your probe generates and again, this is not really where we are going; we want you to be able to look from different places but it's version 0, and it will get there. What I really want to say at this point is that this project only started four?and?a?half months ago and most of the people working on it now only came on board three?and?a?half months ago and in that time, we had custom hardware built, 500 of these things, we have firmware made for that and we have an infrastructure that collects the measurements and can configure these things set up and the website set up and the people who did this worked very, very hard and I think they deserve a round of applause. It's Andre, Anthony, Victor and the leader, Rahoud, and I think they did a tremendous job and I want to thank them.
So, become a probe host, go to atlas.ripe.net and register. We want you to tell us where and if you can tell us which address block you are going to be in and which AS, that will help us also to more evenly distribute these things. Once you have registered you can get your probe at the Atlas desk, it's to the right of the registration desk, the one that has the RIPE Labs poster next to it, and if you have some folks at home who would also like to do this, then you can probably pick up their probes as well if they have registered before. And we will be there the whole week, so don't all crowd around this desk just now, and be a little bit considerate, we have enough probes to go around and we will be there at least until Thursday.
Before I ask for more help, I will show you more video. This is also sort of the intuition that one could do if one had more data like this. This is an animation that was made by Claudio and he has actually worked on sort of geographic visualisations the last couple of months. He will explain how he did this in some other of his ideas at the mat Working Group at 4 o'clock on Thursday, but I just wanted to show you this video to show sort of what kind of things we are envisage to be able to do. What this is, it's a very easy thing, it's DNSMON data for A route servers.net so the content is not really interesting but imagine if we had like 10,000 of these probes and what we could do and not just this one thing.
So what he does is actually he has a number of vantage points in a number of countries and he puts them concentrically around according to the round trip times and you see here that one of his vantage points is very close to one instance of A dot route servers and that is the US one and he moves the countries around based object the round trip times and here you see an instant switch and you can have mouse overs and see what the times are and things like that. This is just an example what have one could do in terms of looking at services but also looking at aggregate data between a geographic areas. That was the interlude.
So, the thing is that, you knows, I have to add mat that 50,000 probes are probably too expensive for the RIPE NCC alone, and so we are looking for people to actually help us finance this, and we have thought about sort of some initial way of doing it, basically saying OK if you sponsor a number of probes and the amounts are from 8 probes to 256 probes, you will get recognition again, you know, the usual thing for sponsorships but you will also get many more credits because for all the probes you sponsor you will get the amount of credits that the host would get so if you sponsor like you know 128 probes, you will get equivalent credits to have 128 vantage points and they don't need to be the probes you sponsor but they can be any of them. So that is the proposition here. We made a pricing decompatible so means €248 but again it's the message is this is cheaper than running your own, than running your own network. At the moment, I am looking for sponsors that are pioneers, that want to be in early, that don't look for SLAs and 50 pages of contracts and all that, those things; obviously, once we get SLAs and once we add features all the sponsors will benefit from them and will get them but at the moment we are looking for people with a vision who would like this to succeed and have some trust in us and give us some money to be credible and to get this going. And if you are interested, see me during the meeting or contact me, my co?order gnats are very well?known.
Where to get more information: There is the Atlas desk, as I said, to the right of the registration desk, it's where RIPE Labs is and Miriam will show you all about RIPE Labs. There are a couple of people around with RIPE Atlas stickers on their badge so feel free to hit them up about it, and Robert will give a longer talk about the technical details and how it all works on ?? at 1,600 on Thursday in the MA T Working Group and we will have in that group at the end, the chairs are organising a session about a bit of how do we further progress measurements in the RIPE area and with the RIPE NCC so that is an interesting Working Group if you are interested in this field.
And that concludes my presentation. And I managed to do 30 slides in 27 minutes, so that wasn't bad. Any questions?
SHANE KERR: This is Shane Kerr. ISC. I have a few questions. First, are you going to publish the source code that is returning on the device?
DANIEL KARRENBERG: Not decided yet. The trade?off here is security through absecurity, and also, well, not decided yet, I am open to input on it.
SHANE KERR: I would like to see it but I understand. I have more questions but ?? the second question I have is about publishing information about what is actually going on in these devices. So, when it becomes flexible then presumably you can have subsets of the entire cloud, I call it a crowd, why not?
DANIEL KARRENBERG: I avoided that.
SHANE KERR: OK. You can have sub sets of the entire group of probes, is it going to be published in realtime, maybe, or after the facts ?? which sets of queries are being run, who they are being run for with the point of this is and things like that ('Cha' what)
SHANE KERR: Gee, I anticipated that question.
SHANE KERR: I am not a plant
SHANE KERR: People believe that. That is what is being asked all the time. Private measurements and privacy of people running measurements and this is something that ?? this is my take on it and it's definitely something that we are going to discuss in the mat Working Group and with the community in general but one thing that I am pretty firm on we are not offering this as a service for private and confidential measurements; if you want to keep your results secret roll your own because there is no community benefit in it so why would we? But we obviously regular those requirements so (recognise) I think the principle should be and we can discuss those principles, should be this should be a benefit to the community and all the results should benefit the community, also those that individually configured measurements gave, but the devil is always in the details right? So what I can think of is some embargo periods like we do with DNSMON right now, aggregation, so you aggregate statistically and numerisation and all these other things. And then also you can discuss who could see what and these kind of things. That is going to be difficult, I think. But not insurmountable at all.
SHANE KERR: OK, that sounds quite reasonable. The final thing, and then I am done, sorry, is the thing that immediately came to my mind when you were talking about realtime status of the networks and things is that this could be very interesting for bad people, too.
DANIEL KARRENBERG: Yes. As it always is, and we are dealing with that as, you know, in DNSMON with the delay (a delay), and we are probably going to have to do similar things here. On the other hand, how far do you want to go? We could do similar things that we did in DNSMON, if you subscribe you get less of a delay, things like that, so that we know the people who actually are getting the results but beyond a certain size of a group it's not going to work because they are going to pass it on.
SHANE KERR: Yes, I don't mean that this is something that shouldn't be done or that something you need to worry about. I think my recommendation would be to do an explicit security audit and think like a hacker.
DANIEL KARRENBERG: That kind of stuff we are going to do anyway because if we don't that is giant.net. But the other thing is, also, what sort of I have been thinking about, if I was a bad guy, and I wanted to do these kind of measurements, they are readily available amounts of available hosts, let's put it that way, in a I could use to do that. So I don't think it's as bad as people, it may appear at first glance.
Richard Barns, mat co?chair and pre registered for two probes already. The general observation I wanted to make here is that that is really good effort to start opening up the measurement efforts here but it still has aspects of central control and I think it will help to scale things if we can remove it to the degrees possible so carry on what Shane was saying about let's open up the software for this, I would say open up the hardware for this and publish the schematics for what you are doing so people know what is running in their data centre. I would encourage that as a general theme. One specific question: What is the plan for sharing data here? I mean what sort of interfaces can people get to see the data that is being collected?
DANIEL KARRENBERG: That is still pretty much up in the air. Definitely, what I would like to see, and it's again open for discussion, is a realtime useable access method to these things. I don't think it's the right way to close something like this and say, hey, you better use our analysis tools because they are the best that were ever invented. I don't think that is the way to do it. I think the good way is to publish the raw data in some shape or form as realtime as possible and have a good healthy competition about visualisation and analysis. Again, early days can be discussed very openly.
AUDIENCE SPEAKER: Just a question, is there a limit of probes run by an individual or a company?
DANIEL KARRENBERG: At the moment there is a practical limit. We only have 500 of them right now and we want to be fair. If a company or individual were to come up to me and say "I would like to sponsor so?and?so many," I would make sure they get them as soon as possible.
GEOFF HOUSTON: Geoff Houston. I wanted to thank you for this particular initiative. As someone who has worked in the operations area of networking for some time, never underestimate the power of ping and trace route. Honestly, it's what keep networks running. What I have seen over the last 15 years is I object creasing amount of private use of that data and we know less and less about the network that we actually all live in and that seems to me to be an awfully bad situation. I think what you are doing here is a very, very good initiative and I would encourage everyone here to think about sponsoring this. We need this kind of information out in the open where we can all use it because quite frankly because this is the data that we build tomorrow's network with so thank you.
SHANE KERR: Thank you very much Geoff. Can we arrange a meeting with Paul Wilson?
AUDIENCE SPEAKER: Mike Hughes, concerned Internet citizen. Following on from Shane's comment, have ?? it's a great initiative by the way but have the devices been subject to any form of security or pen testing yet?
DANIEL KARRENBERG: Not yet but here is some of the answers: So why do with do a hardware and not a software probe. Basically, one reason is comparable and reliable methods, they are temper resistant so nobody can tamper with the results even. Of course they can but resistant as I said. But also there is (tamper) a security aspect here. I don't think that really an attractive or easy target. (They are). And more importantly, other than signature at home and all these kind of things they are not introducing potentially weaknesses into the host systems, they are something closed.
AUDIENCE SPEAKER: Understood, yes.
DANIEL KARRENBERG: If you ?? I promise not to go into the detail details because Paul will kill me if I start that (technical) come along on Thursday. These do not expose any open pours. They are not passive. (Ports). They just connect once securely to the infrastructure and it's all they do other than the measurement traffic and all the infrastructure behind it has actually been designed with security in mind, very much in mind. So, you know, come on Thursday or talk to Robert, he is the resident expert, or to me.
AUDIENCE SPEAKER: Just to close that out. That is great. If you are going to open up the source I would suggest, and release that, before do you that please get some other party to go through it.
DANIEL KARRENBERG: Most definitely, most definitely.
John: Hi, John quarter man,
DANIEL KARRENBERG: He is the guy who came up with the name "RIPE" by the way.
John: Just wish I'd trade marked it. Every penny I might have gotten from these conferences. Anyway, as Daniel knows, Internet perils does something similar, we use very, very few probes and go to a largish number of destinations covering much of the Internet. But I have a question ?? well, compliments, first: Great idea. Hope it works. The question; could you compare this to ARBOR's Atlas project and two points: One is ARBOR of course is limited by where they have their probes as will you be to some extent except unlike them, you are not just intercepting traffic you are actually sending probes out, but of course the more you do that, the more it starts to ?? people will per sees it as disturbing the network which is why I understand why people won't open source for this thing but how many organisations do you actually ? doing what he is talking about here, I don't mean hosting them but orbing it? Probably not very many. My final comment: I am really looking forward to the hard part which is not the probes it's really what you do with the masses of the data.
DANIEL KARRENBERG: OK, well, I am trying to distill the questions out of what you just said. About how many, I still think the more the merrier. About disturbing the network with the measurement traffic, I think, I wish we could generate enough measurement traffic to disturb the network. I think these days are long passed. What was the other question? The perception of disturbance. Well, we will have a black list so if you don't want to be /PROEPBDZ, you are not going to receive any traffic, but at the NCC we have very consciously said we are not going to do passive measurements because of all the box that you open with privacy and all that kind of stuff so we are not in that business, and how to to compare it with ARBOR, what ARBOR does is mostly BGP?based and as ran Dee tells us every time he opens his mouth is like, you know, to make up inferences from BGP data to packet flows is sometimes a little bit farfetched so what weigh want to do here is only measure packet flows and yes ARBOR has some passive things and some even BP I probes but I think we are in a totally different league and my apologies to ARBOR about the name clash; we had it all printed when we realised.
Lorenzo: Internet user. I registered for one at home. I think this is great, I think it will really improve or has the potential to improve latency in the Internet so thank you for thinking of this and somehow it's one of those things that why didn't anyone think of it before? One comment on security, if I were a black Kath I would probably go for some widely distributed system with a powerful Ethernet card and actually capable of running software and running spam than something like this which only has few thousand hosts so I wouldn't worry about security too much.
DANIEL KARRENBERG: This is my prepared answer to the BotNet question. I think we have run out of questions. Thank you very much for your patience and I hope I have convinced at least some of you to participate in this, either as a host or more preferably, as a sponsor. Thank you very much.
PAUL RENDEK: Thanks Daniel. Thank you. OK, great. So nobody rush out to get your probes, the labs desk will be there for a few more days. Thanks again Daniel. Next up, we have a talk from Maarten Botterman on the IPv6 monitoring survey that he has done in 2010. He is here from GNKS Consult. Maarten.
Maarten Botterman: Thank you for that, I see BotNet here on my screen, that means somebody else is moving my slides. Thanks for having me here and after this, I think actually initiative will measuring in the network, one thing you can't measure and that is what people think and business plan. You have to ask. Had a that is what we have done again this year in June. Basically, the aim of the IPv6 deployment is to get the best possible view of the present penetration, the perceived bottlenecks and the future plans. And really by asking the RIR communities. It started I think in 2008 with ARIN, the first initial pilot on such a survey, last year we did it with APNIC and RIPE in the summer, and this year, in fact, all RIRs apartmented so that means that, next year, if you do the same thing again we will have a global overview. Now, this year, and also due to the fact we have got limited speaking time, I will just focus on the comparison of the data between RIPE last year and RIPE next year, or RIPE this year.
It's sponsored so far, the set?up of this, by the European Commission and it's funded basis in the European action plan for IP version 6 which is basically a plan to prevent IPv4 shortages to lead to problems, so slow down of market, to distortion and to negative effect on innovation and the Commission is not done with that. So, if you look at the global participation, what you see is that most Lee ?? the biggest part was RIPE NCC but for sure, at our RIRs have been picking this up and almost 1,600 responses from 140 countries. Half of the country of the 140, had one, two or three Respondents, so it's not representative per country, but 1,600 is representative for the world overall.
RIPE NCC 769 Respondents. Participation, largely comparable to last year, both in terms of distribution around countries in Europe, distribution across categories, about two?thirds of the Respondents is ISP. Some of the responses are only answered by ISPs, others by the full set of people in the RIPE NCC network.
By the way, you will find the presentation including more slides that I hide for the time but if you download it from the agenda, you will find them all. So in general what we see is more customers to use IPv6 connectivity. It's slightly examining up. More ISPs to consider promoting IPv6 to their customers. (Going) and again, there is a pretty steep one, whereas more than 40 percent in 2009 still didn't think of it, it's only 9 percent now. And more organisations have or do consider having it. The A people that thinks getting away without IPv6 is decreasing. Please note this is not only ISPs but also the other users.
So, more organisations with IPv6 in production, overall, but also the percentage of the IPv6 traffic is growing.
Now, how to get many things one picture and basically what you see on the side is all different elements from IPv6 can be used. In thousand 9 we didn't have the ISP to consumers yet and ISP to companies, which you can also find in the latest updated web presentation. What you do see is that the organisation with no plan in 2009 is about half, or a little bit more than half. In 2010 it's really getting to a clear minority of thinking of the plans.
Now, in terms of attitude, perception, experience, you see that, still, the business need not seeing the business need is the main reason for organisations to not to go for the IPv6, but (experience) it's slightly getting better. To go for it, is increasingly not a surprise, the availability of IPv4 address space decline and if you look to the biggest hurdles, the upper set is about organisations who did implement or were considered to implement IPv6. Then, the support has become a bigger issue there. If you look to those not ready for IPv6 stilt main reason is their perceived cost part, and, in addition, you see that vendor support concerns have gone up since last year dramatically.
Interviews in a way confirm this. There is a lot of vendor?related issues, if you talk to people in?depth about the issues they run into, once they start implementing, the equipment is just not exactly cog what is needed, the general available equipment doesn't support it yet or if it does (do) it doesn't always work optimal. There is a lot of balancing that needs to be done, still.
Now, when in production, the biggest problem is still lack of user demand, but we see that the budget issues which in 2009 were the red 17 percent, have now gone up, thanks to the feedback of people, two questions, issues of 19 percent for those people financially responsible don't get it. And the 10 percent is no access to investment money, just the scarcity of resources. So that has gone up.
Implementation of IPv6. What did people say about that? Well, what you can see is that the amount of organisations that still doesn't have it is relatively the same, yet big growth of those who do do them both internally and externally. In a way you see that organisations started to learn last year are really expanding their experience this year. Now, if you look to the IPv6 set?up, first it's two bars and it's the number of ISPs responding, you you see there is more who have been responding that they do this stuff. In addition, you see that dual stack has increased from 88 to 94 percent. Same here for the nature, native IPv6 has grown in dominance, and if you put these things about how people really gear up for being ready to support it, making it way available to use measurements and there is other data out there, what did was measure in the networks the use which used ?? was about 2 percent when they started measuring in their network a year ago and it has gone up slowly about 0.6 percent and in a way, if you look to other data in the field, this is not so different; use has gone up but preparedness seems to have gone up much more. Same for websites. If you look to top 500 websites reachable on IPv6 in some countries you see a huge improvement, like in Spain, Portugal, in Slovakia, but in most you see it's not hit the big things yet.
Headline conclusions for all this.
First, experience has grown, really. It's been picking up dramatically. If you look to the global data, I think also (I look) we see that the RIPE area is slightly leading at the moment in picking up the experience.
For sure, no doubt about it, ISPs take IPv6 more seriously in 2010 and it wouldn't surprise me if we see this is even more so next year.
On the choice, IPv6 or not, you see that the lack of business reason is less the argument now, and it is become clear also to people outside of this room that IPv4 will run out. Lack of vendor support is considered more of a hurdle and whethers it's because vendor support is less well or people become more confronted with it, my guess is the latter. /SOP and problems from relating to the budget has grown, not only because it becomes a critical thing for an organisation so you get to do with people who are not just technical but think strategic and have to decide about budgets but don't know too much about what is underneath, and still some investment capital scarcity issues.
Now, set?up of IPv6 is just a line that started last year which you see is getting stronger, on both networks, internally and externally, increasingly dual stack native, no doubt about it. So with all this, overall conclusion from the 2010 measurements as compared to 2009, is experience with IPv6 has grown a lot, but use hasn't grown that much yet. I think preparedness has advanced so once used will come up it will be easier to cope with that. And the second thing is last year we were really looking at ISPs, hey it has to start somewhere, it's chicken and egg but maybe ISPs is the one pushing. I think the ball has now been rolling towards the vendors more than before. And again, worldwide respondent enters indicated they would like to see this again next year to get the value and use three bars instead of two bars. I think it's an excellent way of getting a feel for how the field moves because preparedness is really what we are concerned about. We need to be ready to address the huge question when it's there. It's not about the use needs to be up very high, (use not huge).
The European Commission isn't done with it, either. Sponsorship will stop for this project, yet the Commission will continue to develop other activities and they do want to work with the community wherever the community finds that useful. It's open to the voice of the community to say, hey, where would you want us to put or money and our efforts. I think there is a high interest at the moment at trials and because skill trials really get things done, to get it off the ground but also to get the results and the lessons from those trials out. And this is something that I must say the European Commission is a bureaucratic organisation, a government, there is many others and it took them a while as well but they got it done as well, they took the effort of launching IPv6 into their organisation as well so I would almost say that deserves applause.
I see somebody from the European Commission smile. Very last one: On the 14th of December, they want to discuss with the community the way forward in Ghent, and you are all invited to that.
AUDIENCE SPEAKER: From French ? I have two more comments or two suggestions for the next study, in the study you don't make really the difference between a ? and a fixed or cable operator. One thing is sure, the path for going to IPv6 is totally different, OK? What we can observe is what fixed operator or cable operator are twisting in IPv6 but mobile operator is still tie immediate, so it could be twisting to separate both and the second one is we cannot see if, for the ISPs that go to IPv6, it could be twisting to make the difference between you do the part, the application level and the infrastructure level, for example some operators, today, are looking for IPv6 for the, more for the application level for the terminals because lack of atresses you mention before but infrastructure is seen as a next step and for some others what they have before ? by your infrastructure but waiting for the applications, so it is two different ways for going to IPv6, it could be reflecting the study if it's possible.
MAARTEN BOTTERMAN: Thank you very much for that, and, for sure, the mobile versions, fixed, is something of interest, it's not covered by the study, it's just focusing on the general and the other one is I think in a way what we see last year ISPs, this year towards vendors more emphasis, the same pressure may be higher in the physical networks or implications as well. Thank you for the input.
AUDIENCE SPEAKER: Kurtis Lindqvist: On the slide you had on the statistics for IPv6 content in Europe I done a similar study top 500, 4 out of 10 are v6 reachable and I wasn't surprised. It turns out it wasn't that good news, I am part of the Google v6 test bed and all those 4 are Google, so I was wondering for the content you lookedality do you know how much of that is unique content or how much is actually Google, of the 500 slides?
MAARTEN BOTTERMAN: I don't know but we can find out, I can ask N NL but it's really looking at the top 500 websites and to pull it.
KURTIS LINDQVIST: The content I looked at is quite tragic, for me it works but for normal user none of the top 10 was v6. It's changing for Norway.
MAARTEN BOTTERMAN: Last year in the whole of European in the top 30 there was one, in the top 30 websites of European countries there is now already almost 20, of 27 countries, so it's very low, still. Yes.
Susan wolf: DNS organise and ISC. The question that comes to mind here is do you think your survey population and your funding agencies would be interested in doing this more often than annually? Because things are going to start changing really fast out there and I'd be curious to see data of a higher granularity than annual.
MAARTEN BOTTERMAN: Thank you, that is a good question to NRO people who are here. It's not up to me but the difference between last year and this year is right, but if we should go to six months from now, I don't know. For sure, it will be an annual finger on the pulse.
Susan: I see some of your people involved in the funding in the room and I am happy to tell them it should be more often than that.
PAUL RENDEK: I am taking that...
Sorry guys, we have got to keep these really fast.
LORENZO: Google. Big content has serious problems in IPv6 today because of persistent brokenness and in home routers, something that we cannot fix and the US manufacturers need to be involved. My contents other big content providers say this is absolutely not an option, they would rather do carrier great Nap than turn on v6 unless the SLA gets better so that is something we need to fix as an industry. I am going to be hopefully talking about it later this week, giving some examples and talking about the ideas we have but this is something that we need to fix together.
AUDIENCE SPEAKER: In the slide where you had these bars going left to right that is the difference between 2009 and 10, you had one bar was IPv6 to customers by ISPs, something like that: That one. There was one thing that struck me as very strange, probably you have a good explanation: The last bar seems to be shorter on the blue one deployed in 2009, is there a regression or the size of the sample change, what is the explanation there?
MAARTEN BOTTERMAN: No ISP to consumers isn't up there. What is going wrong in this presentation which is right on my computer is you get names for all the bars. ISP for consumers is now and ISPs.
ISPs for businesses is under a there and I don't know why it fell away. It's more blue and I think this is reflecting reality as to the experience in talks about organisations, as well, the major Dutch, I am from Holland, operators KPN, UPC, they foresee launching IPv6 for businesses the year before IPv6 for consumers and it's coming up.
AUDIENCE SPEAKER: OK, thank you.
PAUL RENDEK: OK, thank you very much. Thank you. I would also like to thank my other colleagues at the RIRs because we do coordinate this as an NRO effort this year, so it was quite licely done. I had one comment to add: Is Jack about a bow here from the Commission. I wanted to know if you had feedback that you wanted to give to the Commission on what they should be doing, spending their money, what the Commission should be doing to speak to this issue, where can everybody send their comments to. Do you have anything here or that we can put up somewhere where people can mail to that. Would be quite useful if anybody wants to give some comment on what you should be doing with all that cash. There is an address at the bottom is there?
MAARTEN BOTTERMAN: I think if you go to this website, you will find a link to your website as well, Joaques?
PAUL RENDEK: They can leave comments there as well.
MAARTEN BOTTERMAN: Otherwise you Google IPv6 European Commission, and you get there as well.
PAUL RENDEK: If you have troubles right to me, render [at] ripe [dot] net. And I will make sure he gets this on your behalf. Next up we have the RIR panel so if I can ask the representative, John Curran, Geoff Houston, my boss Axel Pawlik and Sergio Rojas. If you can join the panel. What we have decided to do this time, we have asked them to all come up here, give us a few, probably 3 to 5 issues that they are dealing with in their registry and open up the floor for discussions for to you ask any one of them questions about what is going on in the RIRs. I believe my colleague from AfriNIC is not here, I don't see him in the room. So this is the panel that we currently have. I think we will start with Geoff from APNIC. The floor will be for you, if you can just please upload Geoff's stuff. Thank you.
GEOFF HOUSTON: I might as well go to the podium. My name is Paul Wilson [sic] I am from APNIC. I do impressions. This is the APNIC update. I have been asked to change things around from the normal presentation and do things in terms of just looking at current issues. So I am going to give autopsy personal view of the hot topics at A N nick, I am not sure that Paul would necessarily agree with it but the first hot topic that sits in my head is, over the last couple of years, the amount of demand for addresses in the Asia Pacific region has out stripped most of the rest of the world. We are now seeing a number of economies in that region, particularly China, but also career I can't, Japan and India, do massive deployments of both WIDE and wireless. Currently, this year, we are on track to getting through six /8s so that is a /8, 60 million addresses, 70 million addresses every two months. That is the speeds they are going out the door. So that curly line on the right shows you the amount of addresses per day, if you will. And as you notice over the last couple of years the deployment at APNIC is really accelerating, seven out of ten addresses head into the Asia Pacific region so APNIC is doing around /# 0 percent of the allocation rate. When we get close to the bottom of the barrel, as we will do next year, it's APNIC that looks highly, highly likely to be the first RIR to run dry and it will be happening probably within eleven to 12 months from today, but don't worry, you guys will be right behind us.
The second thing we have been doing which is kind of new: Address this year we got given network 1 and that was bizarre. We have this relationship with RIPE where we get RIPE to advertise a few prefixes for us so we get them one dot one dot one dot one and on their 10 megabit ?? we think that is business curious how much traffic does network one get, the answer is a bucket?load. Cool. As we got more /8s and we have got six this year, we started testing all of the others. Not every address is the same, and this is actually a snapshot of network 42, and I was surprised, I was actually thinking that some bright spark at had used 42 dot 42 dot 42. No, I don't. There is a line up there that consistently gets around a megabit per second and for some reason someone likes 41 dot 57 and it's from Europe. One of you. What we are doing right now is testing their prefixes before we start allocating them because we don't want to saddle some poor end user with a massive amount of incoming traffic, with the assistance of NTT in Japan and merit in the US a couple of weeks of incentive testing of every /8 before we let it out for allocation and this is a /STPHAZ' a multi?coloured picture. The next hot topic, we have been working very hard on RPKI and resource allocation. The outcome of that is we are now a production service at APNIC and in our case users of the portal there are able to generate full flown certification of resources and routing origination at our stations. That is part of a RIR wide effort to coordinate across all the RIRs to get to this in the coming months and we are working very hard with other implementers to make sure the products of this are interoperable. As we get into areas of address shortage certification of your addresses and the uniqueness of your addresses, will become increasingly important and we would like to think this is just part of building some foundations that will see us through the coming years with a little more surety than otherwise. So again a hot topic.
Last but not least when we went to the meeting in Beijing which is now about a year or so ago, our members said your procedures are completely stuffed and make this too complicated, make them easier. We couldn't resist this so we have now done a project to make obtaining IPv6 addresses a lot easier.
If you have already an IPv4 allocation from APNIC and you are a current member, obtaining an equivalent IPv6 allocation is as easy as one click away now. So we have gone very down into this complete simplification, the registration APNIC is not the reason why you are not doing v6. Thank you.
I will relay that back to Paul, I think he would be gratified, we did spend a lot of time making it very simple. That is our hot topics. We believe discussing all of this and more at the next APNIC meeting which is going to be held in the Hong Kong special administrative region, wherever that is is in China at the end of February, where the weather will only be slightly warmer than it is here in sunny Rome, if you like this kind of weather and that kind of venue, please come and join us. Thank you.
JOHN CURRAN: So they wanted to do the panel format, I told Paul I would do 71 slides and talk really slow. But for sake of everyone I guess I can't really do that, so let's see, just the important stuff:
We are continuing to work on development of our web?based system and ARIN was a little later than other folks to develop that so we have a very big development road map and making progress, ARIN on?line exists today, you can do a lot of updates you used to be able to do manually, can now all be done on?line. You can see the road map on?line, but we are very, very happy with the progress we are making. We are up to, I would say, up to mostly most of the other RIRs at this point, now looking at some of the more advanced automation. Outreach on v4 and 6. We are also looking at RPKI services and as Geoff said; the APNIC is certainly wildly ahead here. In the case of the ARIN region, one of the things New York America has is a lot of lawyers and so when we look at new services we have to often engage them. Steve isn't with me here on this particular trip but that may be the snag for ARIN in getting RPKI getting to the point, we are trying to do everything we can to move expeditiously production to that stage. I guess the most important thing as we move ahead is we are also continuing our Whois services, we are busy, we have moved to a restful interface which allows direct automation get the ARIN database. Some people realise /SWEUPS are going away and when they do they want to take the current back end systems they have that send e?mail and have something more a little less Neanderthal.
I will skip some of this for the sake of efficiency. POC validation. Last ?? or ARIN region said we have a lot of old records, ARIN is the successor registry for all the regulatory resources in North America that were signed prior to 1990 ?? or most of 1997 and before and what that means is in a lot of cases these people want to use their resource and be left alone. And that is great, except what it means really left alone, they don't want to bother update any contact information. That has turned out to be a little bit of issue, it means there is resources in use all over New York America that we don't really know the contact for. We have started automated system to contact all of the contacts in the Whois database via electronic mail with one click to acknowledge that your e?mail is current. Sort of like people are seeing if you have a domain name registered. We have sent 4 /# thousand points of contact, e?mail, these are people with direct contacts or association with ARIN, we got about 30 percent response rate. We are actually marking the Whois database in ARIN if you don't respond at all we will mark the contact as not responsive, so you can actually have a circumstance where we have tagged your contact invalid if you don't respond to those e?mails. We are actually going to begin the next step of contacting indirect points of contacts so these all of the resources that were signed via reallocation from an ISP and may or may not have been updated, the goal of corks trying to get current contact information for all the records that are in the database. Big effort you will here more about it. And it has many implications as we start running out of resources.
Policy implementation, I am going to skip that. You know how to get to the ARIN slides.
I am going to talk about new policies, policy proposals, globally coordinated transfer policy, PP 119, effectively any remembers may transfer IPv4 resource addresses to the resource registry of another as long as the two agree and exercise Internet storage ship and. Right now, there is no inter RIR transfer policy, and some people are indicating that might be useful to have particularly as we get towards the end and resources run out. This is a proposal that has just been submitted in the ARIN region. Another one been submitted, protecting number resources. ARIN shall use any reasonable and... obtain or abandoned number resources and seek the return of those resources to ARIN."
ARIN, we have encouraged people who are old address holders to sign a legacy RSA, certain protections, pay about 100 dollars a year, you have a very favourable relationship compared to the standard RSA in terms of cost and application of policy, but if organisations don't sign the RSA we haven't done anything and in particular we don't require the L RSA, we haven't spade to people you have to complete one. This is interesting because this policy would sort of say, you need to show that you are not aabandoned. That means at least verifying your contact and depending on what the region does it could mean signing a regular see resource agreement. They are going to be discussed but you should know about these in the Karen regions because while all the ?? a lot of addresses are presently going to APNIC a lot of them did originally go into North America so these are going to be interesting that could affect availability of v4 resources over the next few years.
And one more and final thing: We are going to be discussing these on?line but also the two upcoming meetings: /SA*PB within a in Philadelphia next year in conjunction with NANOG and will simply be an ARIN meeting. That is it. Thank you very much.
PAUL RENDEK: Thanks, John.
SERGIO ROJAS: Before to start, accept my apologies because my English is not very good. Maybe you will hear some Spanglish in some phrases or if you have some difficult contact with me and I will clarify some difficulties.
Well, I will show you an update of LACNIC. So to be understandable, we divide it into sections, this is we are working and we work during the 2010 and the first section is for internal work and the second is orientated to the client or to the customer.
So for the first point, we are working in project, that name customer orientated, and the goal of, collecting all information, hosting in each computer and put in our central database so in that way we can give to the customer or to the member, correct information, for example when they have some difficulties, some question or they call to know about the status of the resource, for example, we can, any people or any employees of LACNIC can use with the help of some tools or software that we develop, give the correct information to the customer.
Well, new website, we are planning to make new website and the release date is December of this year. Op working in policy ?? politic positions, we are collecting policies and they change and alter these policies.
What to make it possible, we are working in the system for engineering. It is with this ?? with this point it will possible to make the internal procedures and the external procedures so we have to buy new servers, switches and maybe develop new software to make it.
For external for or for the customer services, the project RPKI is, well we are finishing our development is (RPKI) working in a module that allow the possibilities to our LIR use on this tool and RPKI is a tool to give to the customer to issue digital certifies for the customer who has Internet resource, for example is the way that they can prove the /RAOEUFT the use of the IPv6 or IPv4 or ASN. The project is not the beautiful reception of LACNIC. Well, for its name is Spanish, (Spanish). If we translate it's automated system top administer resources. As we know, each provider or each customer that reallocate or make a sumication of/29 or less must register in our system to have the information published in the Whois server. So in sometime largest or biggest organisation have their own system to make this registration, so it is hard work or an extra work to keep up?to?date both database, the database from, to keep up?to?date the database from the client or the customer and keep update on the database from LACNIC. So we implement or install a new server that make the possibility to synchronize both database and give the same information in the Whois server.
We are working to DNSSEC and IPv6 project, we are starting to make some serving or contacting with members who has or who hasn't IPv6 for and as ask you know in South America, the IPv6 plug is not very announced yet so we are contact with them to know what can we do for them and how can we help you to announce and to promote this protocol.
Well, here we are ?? here, we are our membership evolution. I won't talk about the evolution but we have, now, 1,000 and 432 members, most of them are in this small category, I mean organisation who is at least a /20 or from /20 to /19.
Well, during the 2010, 10 to some important forum meeting and well here is ?? I don't have time to talk about. Talking about projects, too. As a social responsibility we have three important projects, name a.m. PA R O, FRIDA and RIAI CE S, a.m. par row is a project to create a group (a.m. PA R O) a group on each country of South America, to have an incident response team. So I think we have two or three countries that have a group, a formal group that they are working that and we are trying to join more countries. The project FRIDA, well here gave eight to the little projects or to the organisation or member that wants to implement a new project like IP telephoning or connection in the countryside or if, for example, some organisation or large organisation who has ?? they have sometime little project to make easier or to get easier to the Internet connection. There is some example that, well in Peru or some countries, that they can't get easily Internet connection. So FRIDA give financial help or help in ?? help for this project.
And RIAI CE S is to make copies of route server. There are plans to install in a country in South America (small letters) and I think one country is missing. Well, before ?? we had our second meeting in the same year, it was LACNIC 14, it was in sue Paul low, Brazil. We had almost 300 participants or people who attend. Five policies were presented and four reached consensus. And for the first time we had to present our inventory or our free space or reserved space of IPv4 blog and we present possible date of IPv4 exception ?? exhaust Sean, according to our stock. And to be possible, well our technical manager, well we took on the last 36 months of allocation and assignment, and we dealt with some module, we can ?? well, being some pessimistic, we thought that we will give the last IPv4 in December 2012, but being more optimistic, well we can go even to the May 2014. But this is a lot of studies that (there is) we made, for example, some pollee module or linear module, there is presentation with full information about this on a list analysis, well, if you need it, I can give you.
Well, our next meeting will be in Cancun, Mexico, so you are invited. We will discuss, ask the ?? our pre views meeting, policy forum, technical forum, security, IPv6 and, well why not, sun beach and a lot of fun time.
As this year, we are planning to make the second meeting, too, in the same date, in 2011, but the place is not confirmed.
So thank you very much. As I say, if you have some questions just contact with me.
PAUL RENDEK: Thanks, Sergio. Next up is Axel who will be providing you with a short report from the NRO. After he is ton it, I am expecting questions.
AXEL PAWLIK: Good afternoon, I am Axel Pawlik, I am the chairman of the NRO.
What is the NRO? Basically, don't fear, you know them all, it's just the RIRs RIRs playing together. We decided it would be good if we had a somewhat but not too much formalised vehicle to play vehicle, talk with one voice to the rest of the world. At the time the priorities were, still are, protecting the unallocated address pool, in the case IANA falls into the water, they are in California, after all. You should make safe that you are somewhere else, right? But I mean computers and stuff. It happens.
So, then we would be here in safe ?? here in safe Amsterdam and be able to do the list with the numbers and stuff. Easy enough. Most importantly, we are there to also protect, talk, to promote and further the bottom up industry self regulatory process, you people coming together regularly talking about policies and regulating the way you do business according to numbering resource, for instance. And the last but not least, to save you the way around the world to Cancun and other terrible places where so much sun is not good for your skin, so you can talk to the people in Amsterdam if you had something to bring to all the RIRs together. Of course the other way around if you prefer sun, you can go to APNIC and not come to rainy Amsterdam.
We also at some point in time looked at the way we are working together with ICANN and we have resigned an MoU on the 21st of October, 2004, where basically we say the NRO, now that it is in existence, would like to act as the address supporting organisation within ICANN. We thought that was the way we did it, an improvement over the previous MoU, it's a little bit clearer and I think it works quite well.
Now, as we are informally the RIRs working together, we also decided it would be probably fairest to move the officers throughout the RIRs, so you saw I am chairman for this year, after the end of this year I can retire for a another couple of years before I have to do work in the NRO again. We have currently as secretary, Rahoud, and treasurer John and next year they all move up a little bit and Rajoul will be chairman and John will be secretary and then I don't know who is coming up.
JOHN CURRAN: I can keep being treasurer every year.
AXEL PAWLIK: We have those fights occasionally. We have as a vehicle for more precise coordination, for /PHORPLal, between the RIRs, the coordination groups, we have two of them, the engineering and communications groups and Andrei and Paul are chairing those until the end of the year and I think then, they will be happy to retire.
Speaking about ICANN and, for a very long time we have shared the cost for our ICANN ?? ? ICANN contribution, among the RIS and we have some complicated formula that depends on the amount of addresses allocated, which is a good thing, isn't it? Yes, for us, but not for you, next year.
And also the amount of members we have and we come out with some numbers and overall it works quite well. Also for the last, I don't know, five years or so at least I think, we fix the amount of money we give to ICANN, so that works well.
We do go to ICANN meetings. We tend to have representation at every single ICANN meeting although not every one of us goes to every meeting. There were more than two this year but one is coming up. Anyway. So we had Brussels in summer, where we had a very short meeting with the ICANN CO E, sometimes those guys are very busy similar to us here at RIPE meetings, have to find five minutes to talk to them. Regularly, at ICANN meetings we do consultations or small presentations and chit chats and a bit of a discussion with the governmental advisory committee that falls under the outreach that all the requires are doing together and also individually and we had a short update at the public forum. Now, ICANN hasn't seen much of the RIRs in a very visible way for a couple of years and we thought in Brussels it was probably not bad if we raised our visibility there a little bit in terms of what are we doing with the IPv4 run I couldn't tell that is all over the press? We will be the bad guys in the end anyway so we can just as well stand this and tell them what you are doing and how things are going to be so. We will do that in Colombia a little bit later this year and talk more about this.
The IGF is a governance forum, big workshop, Internet technical community, governments and everything in between basically, base, talking about how these things all work. It is a great thing. And we were slightly concerned when this started five years ago but it turned out to be very, like I said, very beneficial to ?? in helping our outreach to those people. Also, we have a multi?stakeholder advisory group that you can imagine there is a bit of a programme committee with all sorts of people from all walks of life and ? for the RIRs. We had a twist in timing this year so this year's IGF has already happened. That was later in the year. But it happened earlier this time. Again, it was probably the nicest I have seen so far, they missed only one in between. We have gazillions, of workshops ? not gazillions, but probably hundreds, and from you it's expected that you put forward something that you would like to talk about there, and we did a couple of them ourselves and did some participation in our people's workshops that touched upon things that we are walking about regularly. We had a people ?? we had a booth, point of focus they could do to and we didn't have a coffee machine but people came up to pick up leaflets and talk to people and that is good, and, over the course of the event, lots of interviews with journalists and the like. All over. It's a great thing, we support it going forward in the current format. The most important about IGF, it's into the decision?making forum, so that is good, we can go there and relax and talk to our peers and others without being afraid that we have to defend pieces of text and fight for days and nights over little words, which other people do.
Now, ITU issues. Basically, we go to Geneva and to other places, whenner invited, and sometimes even ?? I don't know, we can go when we are not invited because most of us are sector members anyway, we go when we see see something interesting on the agenda, it might be a study group that goes on for about a week and it's difficult to make out when precisely you have to be there for this particular agenda point, but we manage quite well, in general.
Interesting development was this year, the IPv6 group that was set up basically to discuss things, how, IPv6 deployment goes and how we can help everybody to roll it out and to look at problems with the allocation of IPv6 addresses and fix them and we said, oh, problems we don't know about it but if we don't know doesn't mean there are no problems. If we hear about we are eager to fix them. There were a couple of mailing list groups set up, one about development and one at looking at those problems and finding ways to fix them. I am happy to say no concrete problems were found and that particular mailing list was put to sleep in a good way, it's dormant and it might be reopened later on. So we have following all this quite closely (we are).
The ITU has general meeting type of thing where all the members come together like we do on the Wednesday and we complain about that it's maybe two hours or three hours, sometimes. For the ITU those things go on for three weeks, in Mexico, this time. I don't know whether there is so much of the sun there. We thankfully had people going there, not everybody had to go from our part of the world; we wanted to follow this thing and be aware of developments, but the Internet society basically sent ?? sent every night, a bit of a report out to the rest of the Internet community. It was very interesting to follow in wierd way. It's not something that I would recommend to go to or to go through. But looking at it and analysing what has happened from afar, it looks like it went OK. There is nothing too threatening in the letters, but of course letters need to be interpreted and then we will see what happens over the course of the next year or the rest of this year, how these interpretations are being put into practice and again the only thing we can say, we are here to help and don't be afraid of us. Yeah, yeah. Of course.
Right. Other activities, obviously coordination and the engineering field, we have heard from my dear colleagues here that all the RIRs basically are busy on certification, communications is something that we said years ago, we don't want to do any lobbying, we don't do any lobbying but we do a lot of outreach and talking to people and that works quite well, that is /TPOUPB do, mostly, unless it's in the middle of the night.
And we ?? also, it's nice to be able to send e?mails and do things remotely but occasionally it's nicer to sit around the table for a day and just get things done so we are doing this, as well.
And that is basically all I had to say, before I get bashed by Paul. Thank you.
PAUL RENDEK: Thank you. OK. So, if there are any questions, then please fire away, we have all these panelists here from all the RIRs. Ask away.
WILIFRIED WOEBER: One of the chairs of the RIPE database Working Group. I have got a question to John with regard to the mass mailing success, I would rather labour it like that. You said you had 30 percent of replies. That is less than 50%. So my question is actually, what do you plan to do with the remaining more than 30?
JOHN CURRAN: So, at present, the POC validation, or point of contact validation, we sent multiple e?mails to the address for POC and if we don't receive any response after the third e?mail we get 60 days to it be acted on. At that point we mark the POC as invalid. Now, the only thing that is happening at present is that a record that has all invalid points of contact there is currently a proposal that says that should be reclaimed, and that is sort of an interesting one. There is also, ARIN will be sending out shortly for community consultation, a change that says if you are an invalid point of contact you can't make up dates in an ARIN on?line because, at appreciate, you can actually, you go in and you do updates and similar even though you haven't acknowledged that you are a valid contact. We want to know if you are invalid maybe you are and we shouldn't be working with you. So it's up to the community what it wants to do. At present, all we are doing is tagging contacts that don't match validation. There is also a bulk file available on?line if you subscribe to the bulk Whois system where you can download every point of contact, every resource that has no valid point of contacts. Some organisations that are involved in things like anti?spam and similar, seem to find interest in that bulk Whois file but unless the community directs us to do something else, we are just going to mark them "invalid "potentially preclude them from using ARIN on?line until they click, submit and say they are a valid e?mail address.
AUDIENCE SPEAKER: I have a question about depletion. /AP I can will go first, most likely and then I see the other regions like 2014 and so on. What I was wondering about this, is it possible for the demand to shift from the APNIC region to the other regions by simply registering LIR in the other regions and simply getting the addresses you need from another region and take it back to your own country and use it there?
GEOFF HOUSTON: Generally, the RIRs service a region and they generally prefer the demonstration of need to be based around network deployments within that region and certainly in the past where we were seeing global enterprises they have taken memberships in awful the RIRs and worked in each region with each RIR. However it's true to say the world has an awful lot of folk who are big multi?national and it's also true so say we are currently consuming more IPv4 addresses now than ever before. Such a demand has its own impetus and dough men you tum and there are a lot of inventive folk out there, the response I have given you is certainly our general intent but it may well be we do see some amount of transformation of demand into other RIRs to the extent that folk are meant to demonstrate renalinal demand that is what we foresee. To the human inventtiveness is a wonderful thing.
JOHN CURRAN: I would like to follow up on that, so ARIN also, we service service region and expect organisations to be documenting their need for address space within that region, for resources to be used within that region [mark]. It is possible, I have run several ISPs, it is possible to take demand from a number of resources in one region and solve it somehow and make a shortage appear in another region, it's not too hard to some limited extent. And that is fine, it's human inventtiveness as Geoff calls it. But someone would actually wholesale creates demand that doesn't exist or creates illusionary infrastructure in a region that suddenly needs numbers, in the ARIN region that is fraud and you can go look at our website. Number of resource fraud results in the resources being zeroed out and revoked and returned to the free pool, so that is something, if you were serving a lot of customers and you wanted to make a fraudulent request you would have to wonder whether you wanted a large list of resources removed from you as a result.
AUDIENCE SPEAKER: My colleague isn't here from AfriNIC on the panel (Paul) I can say there is a policy proposal in AfriNIC a fair run out because at the end of this whole game, when the rest registries have run out of v4, AfriNIC will have an ample amount probably for the next two years who knows, maybe 4, Geoff is saying four, I am being a bit conservative here, they will be recording in that policy what would be used inside the region will be taken out and they would try to stop any of that as best they could. So that is what AfriNIC is up to.
SERGIO ROJAS: Just to add a comment and these date and that I present is according to the current policies so the policy can change. So this date is ?? (that was Sergio).
AUDIENCE SPEAKER: I am from the Netherlands. I have got a question for the AfriNIC. I personally originate from China and my company in China is one of the ISPs but we have no way to getting the IP resources from IP nick because ? in my nation and we don't know the knowledge and we like to know how you are distributing the IP to the Chinese nations and how we, in China, to become that part of the community?
GEOFF HOUSTON: In the APNIC region, and indeed in part of the LACNIC region, we have these things called national Internet registries where a national community has its own registry infrastructure and does distribution. Part of the rationale for that goes back to the origins of APNIC where folk were saying you are doing dealing in English, we don't speak English, we speak Japanese, Chinese, Korean whatever and the case was made into the community that these national registries could better service their own national communities in their own time zone and language, so we have national Internet registry in China, in Korea, in Japan, in Vietnam, in Indonesia, I think that is all. And certainly, as part of this, in China. So the general referral would be for China that there is an organisation called CN nick, the Chinese national information centre that does distribution of national number resources in China. It's not exclusive. If you want to go directly to APNIC of course we will also work happily with you. Randy Bush wants to add something here.
AUDIENCE SPEAKER: OK so it's possible for us directly to go to APNIC and request resource.
GEOFF HOUSTON: That was always the deal with the /TPH*EURS that any organisation within that area could come to APNIC if they so chose.
RANDY BUSH: I O J. Surnet, for instance, allocate ?? has an allocation directly from APNIC. CS T net goes through CN nick. I just spent two weeks in Beijing. There are nobody ?? and C ?? there is nobody speaking within the country of problems of getting IP address space if you are having problems getting IP address space, I will gadly introduce you to the wonderful people I met back there who will help you in your native language, as Geoff says.
AUDIENCE SPEAKER: The language ?? it's European meeting and I believe not ?? 90 percent of us wouldn't really speak English at first language here, as well as in Asia a lot of people do speak Asia but the problem, we do get to get IP direct from your organisation but thank you for answering.
GEOFF HOUSTON: Thank you.
PAUL RENDEK: Any other questions for the panelists here? I just wanted to let you know that we will look like a very different kettle of fish in May when we come together again and you have got these people all standing here. Aren't you interested to ask them any questions as to how that is all going to look. Just in general. OK. All right. Thank you. And thanks very much to everybody for that.
So now I should be letting you go but I am not actually because I have two very quick things that I'd like to do before I go. I ask like to ask my colleague, Andrea Cima, to come up and give us a few figures on what is going on across the board with the RIRs and the numbers. We need to have the elections for the ASO ACCed that has come up and I will ask the Chairman, Rob Blokzijl, to come up and he will lead us through that.
ANDREA CIMA: I will keep it short. My name is Andrea Cima, registration services, and today I will present you with the number of Internet status report. This report is put together by the five RIRs and it simply shows the resources that have been received by the RIRs and the resources that have been issued by the RIRs. The date, this report is as 30th September 2010 so some things have changed in the meantime.
If you look at the global pool of IPv4 address space we can see the green bit of the pie, 35 /8s are reserved by the technical community and therefore not being able to be issued to the RIR.
The great bit, central registry, shows 91 /8s. These are /8s which have been issued to organisations before the RIR system was actually set up.
Then, below, we can see that the RIRs have received 116 /8s. As I said before, this presentation has been put together at the end of September; therefore, this number has changed because in the meantime APNIC has received additional two /8s and AfriNIC an additional one bringing this number to 119.
If you look a little bit more, APNIC 4 /# /8s, ARIN 3 /#, AfriNIC, 4, LACNIC 8 and RIPE NCC 42.
Now, this leaves the pool available for allocation by the IANA to eleven /8s, so eleven are still available for allocation.
Now, how much address space has each RIR issued to its members, as Geoff Houston has said previously, we have seen, over time, a continuous growth and this is reflected in this graph, as well. As you can see since 2008 the amount of address space issued by ARIN has slightly gone down. In the RIPE region we are kind of stable at a little bit less than three /8s and the region that has the biggest growth over the last few years is ? with a little less than six /8s [mark].
If we look at the total amount of IPv4 address space issued, we can see the RIPE NCC has allocated about 28 /8s. APNIC, about 36; AfriNIC, 1.6; LACNIC, 5 /#?RBGS, and ARIN a little bit more than twenty?five /8s.
Now, moving to AS numbers, also here it's quite interesting to see the change in trend over time. If we look at the period from 1999 to 2004, the ARIN was the RIR assigning the highest amount of AS numbers. This role has been taken over since 2005 by the RIPE NCC and as you can see, until now the RIPE NCC is the region that is assigning the highest number of AS numbers. Now, what is this due to because it's an interesting figure. The main reason is that in the RIPE region, we have made, for example, this year, over 2000 PI, provider independenter, asites. These are blocks of resource which make an organisation independent from its upstream provider and these don't only want independent IP addresses, but also want independent AS numbers so that is why the number of AS numbers is so high.
For the rest we can see a stable number, in APNIC, LACNIC and AfriNIC, over time.
Looking at the total amount of AS numbers issued, it's interesting to see how ARIN and RIPE NCC have assigned almost the same amount of AS numbers over time, APNIC a bit more than 6,000, LACNIC around 1700 and AfriNIC at around 600 AS numbers.
Now, we have introduced a new slide which is the 4?byte AS number assignment to see what the trends are with this AS number over time. We can see between around 2007 and 2008, the only require making 4 byte AS number assignments has been APNIC ?? successful in making them. In 2009 and 2010 we can see the RIPE NCC has taken over this role with assigning about 4004 byte AS numbers this year only. And we can see that LACNIC also has taken a stable amount of AS assignment numbers of this type.
If we look at the total A4 byte AS numbers assigned RIPE NCC has assigned a bit more than 600, AfriNIC, 14, after lack lick about 150 [mark].
Now moving ton to IPv6 address space. If we look at the doughnut on the top left?hand, the dark grey one, we can see all the IPv6 address space available. Out of this, a/3 has been got and dedicated to global unicast. We can see that the IANA reserve contains 5 had 06 /12s at this point and each of the RIRs has received in 2006 a /12 from which they make IPv6 allocations. Now, there is another /12, you can see the little round doughnut called miscellaneous which contains some reservations of IPv6 address space and the/23 which were handed over to the RIRs before October 2006.
Now, if we look at the time how many allocations have the RIRs immediate to their members, we can see see here there has been stable growth over time, that is actually good to see. We can see that the RIPE NCC has made a big jump in 2008 with a number of allocations, we can see it in green. Now what is this due? It's mainly due policy change which was implemented between 2007 and 2008, that removed the requirement for a member, for an ISP to plan 200 IPv6 customers within two years. Taking away these requirements using the approach and let's say boosted the confidence of LIRs and amount of allocations made has risen allot. We can see in 2010 the number of IPv6 as made by incompetent nick has grown a lot, mostly due policy change where APNIC members which have IPv4 address space qualify a.m.ly also for IPv6 so we can see the strong impact that the policy making process has on the resource distribution.
If we look at on the left?hand side we see the total number of allocation made by the RIRs. On the right?hand side the amount of space that has been issued. Now it's interesting to say that how especially RIPE NCC, APNIC and LACNIC have issued allocations which are much larger than the minimum allocation size which is a /32.
And now for a last slide and I will be done. IPv6 assignments to end users. We can also see this goes back not too many years, 2002. And has mainly seen ARIN issuing IPv6 assignments to end users. Over the years in the RIPE region the policy for providing independent assignments of IPv6 is quite recent and that is why we can see only 2009 the RIPE NCC has started making these assignments and it's interesting to see that we have made about 200 so far since the policy was implemented.
On the NRO page you can find a copy of the presentation and all the statistics and I don't know if you have any questions.
PAUL RENDEK: Thank you very much. Any questions for Andrei I can't? OK. Fabulous. Then moving on to the last bit of today before I let you go, I will actually hand this to you.
ROB BLOKZIJL: Right. Last action of today is the election of your representative on the Address Council. Let me recoup rate [sic] what we are going to do now:
You have learned, this afternoon, if you didn't know it already, the NRO has a number of resource council which also acts as the Address Council of ICANN. So, it's two names, one body. The council has two sorts of members: It has members appointed by the boards of the RIRs, and that council one?third of the membership and two?thirds of the members are the elected by the community. That is you. Members are elected or appointed for a three?year period and it is all staggered and so every year we have the responsibility to fill a slot because a slot comes free.
The outgoing member, this time, in ?? for our region, is Hans Peter Holan, who is there, so that is the seat that to be refilled with a candidate.
Now, we have one ?? one slot, we have one candidate, we have had a call for candidates going out, I don't remember but there is an agreed procedure with time lines on calling for candidates for this election, and that has been done, apart from Hans Peter, who has indicated that he is willing to do another term, no candidates have been put forward so we have one slot, one candidate. The ITU planning potentially has been mentioned, a three?week meeting. You know what I did on the first date? I spent the whole first day with a procedure to elect a secretary general, they had one slot and one candidate and they had an agreed procedure that took them a whole day. I think we can do better.
I think in election technology, it is one of the fields where I feel we are much better than the ITU. So what I would like to suggest, and this is based on previous procedures, is we give Hans Peter the floor for one minute so he may introduce himself to those of you who don't know him, if there are such people, and then I will ask for objections to his canned see and if there are not too many I will ask for an overwhelming round of applause and then we declare him elected.
HANS PETER: My name is Hans Peter Holan, I come from Norway, and work for Nordic or northern European ERP Working Group running the IT department and their network. That is my day job. I have been on the Address Council since ICANN was first formed and the Address Council was established, so I'm really looking forward to have another term if this community wants me to do that. I think we are now entering some very interesting times and I really want to fill in my contribution to see us through those times.
ROB BLOKZIJL: Thank you, Hans Peter. I want to remind you that Hans Peter is very modest and so I will complete his statement by saying that he has not been only on this ICANN Address Council since day one; he has also been a very successful chairman of that council. He also been for many, many years a very successful chairman of the RIPE Address Policy Working Group so he knows what he is talking about.
Are there any objections to my proposal to re?elect Hans Peter in this position? 1,2, 3... no objections. I see a thumb and it's a thumb pointing upwards. I see more thumbs. Let's give Hans Peter a round of applause.
And I hereby declare Hans Peter re?elected by consensus, there is this word again, of the RIPE community. Congratulations, Hans Peter.
Hans Peter: Thank you, Rob.
ROB BLOKZIJL: And that is the end of this part of the programme. I think we are running into drinks time already.
PAUL RENDEK: I will close this, and Hans Peter, we look forward to supporting you my colleagues at the RIPE NCC. We are closing this day.
THE PLENARY RIR/NRO REPORTS SESSION THEN CONCLUDED.