1 conference.apnic.net/37/program#session/66323 ************ >>Krystal Waine: Welcome, ladies and gentlemen. This is the last hurrah. We have reached the end of APRICOT, so hopefully you have had a great week, you have met some new people, made some networks. Our first speaker today is Pindar Wong, who is chairman of consultancy at VeriFI Hong Kong Ltd. Pindar will be speaking today about routing money on the Internet. >>Pindar Wong: Apologies to Christian Huitema, but does anybody remember the cover of this book? Good, I can sit down now. This book was an inspiration to me because, like APRICOT, you learn what you need to know for the next year. I happen to have a role in ISP back in 1993 and we didn't know what we were doing. We were figuring it out as we go along, read through ROCs and trying to swap notes at APRICOT, practical history. 2 This book was inspirational to me because it was that's kind of what we're doing. What I was trying to do here was borrow from the title, which was routing in the Internet to routing money in the Internet. The reason being, this is a Closing Plenary, and I think the role is slightly different, which is instead of trying to set the tone for the event, I would like to try and stimulate some thought looking forward to the next year. I just need to do a bit of a disclaimer, because this talk is not about bitcoin. How many of you have heard of bitcoin? How many of you have heard of Mt Gox? Okay, this is impressive, I know I'm in the right place. I'm not going to talk about that, so let's move on. I will move on briefly as an example, but if I have the next slide. The next three slides we are going to play a guessing game, which is which global network is this. It has a history 3 stemming back to 1973 and this is the graph of the number of countries connected to this global network. By 2004, it reached about 200 countries. Similarly, over that period, these are the number of members connected to this global network. That now in about 2013, was about 10,000 members. Last but not least, this is the number of messages that this global network passed over time. It now reaches in the order of 4.5 billion messages per year. There is a bit of a giveaway here. You'll notice in this graph on the right-hand side the bottom one is 2008, 2007, 2008, the graph goes down. Does anyone know what happened in 2008? The global financial crisis. No surprise. This network is a global network which is not the Internet that has also been in existence for 40 years, which is the SWIFT network, which is the Society for Worldwide Internet Bank Financial Telecommunications. 4 The interesting thing is this global payment network, which has been pressed into service over the last 40 years, is present in 212 different countries, 10,000 different banks connect to this network. It's a network that you and I do not have access to. It's a global network with bits on a disk that sit there that is in this banking network with tallies, which is not the Internet. The interesting thing is that this network was connected to Hong Kong in 1980. Contrary to popular belief, it no longer runs on X.25. That was dismantled in 2005. It's interesting, because they also have free and open standards, ISO 20022. Today's talk is really divided into three parts. The first part I'm going to share some experience. I'm from Hong Kong, made in Hong Kong. I'm going to talk to you about some activities that we're doing here in this so-called digital buffer state. I want to talk 5 about tragedy and hope as it relates to the banks in Hong Kong. But I want to mirror that with what I see as tragedy and hope, the writing on the wall for ISPs and I think you all in this room are ISPs. Last but not least, I'll talk about identity service providers and how it related to some of the W3C work that I'm involved with. Hong Kong digital buffer state, we commercially connected to the Internet about 1993 and by 2004, we have wonderful broadband, wonderful wireless data. Are we done? I basically think so. I think our infrastructure as a build-out is basically complete. We have enough bandwidth to do what we need to do and it's all there. But as an open and free economy, what we are trying to do is pivot. We have a deepwater port, lots of physical containers, trading physical goods. What we are trying to do, some of us, is position Hong Kong to trade the next 6 thing, which is the global trade of intangibles, intellectual property on the Internet. It turns out that we have built, over the last year and a half, five different exchanges where you can list your intellectual property and then you can create a market to buy and sell. It turns out that there's a problem and that problem is, if you want to license your photograph or if you want to license your web design, getting paid for it is incredibly hard. It's not hard, but if you are talking in the order of dollars, or even cents, then the cost of routing that so-called money globally becomes prohibitive. Even if, like the phenomenon of iTunes, 99 cents for the licence of a song, there is a whole business dynamic that occurs when you reach the right price point. But routing that 99 cents globally is a problem because one and a half billion people don't have a bank account or don't 7 have access to a credit card or don't have access to the Internet through fixed line. The interesting thing is Hong Kong is also a global financial centre. There's lots of things that we are doing in terms of the Chinese currency, the yuan and how it is going to be internationalized. At some point, China is going to want to license its IP internationally through Hong Kong because we have the so-called rule of law. We are also the second largest trading place for the yuan internationally. Recently, or at least over the last year and a half, we have had lots of interest in cryptographic currencies; bitcoin and the 80 other different variants. Is it really money? The title here is "Routing Money". Some people believe so. For those of you who know bitcoin and been following Mt Gox, mistakes have been made. Just to give you an idea of the banking sector, we have one of the 8 world's highest concentrations of banks on the planet. Out of the top 100 banks by market cap, 70 of them have presence in Hong Kong and we have a three tiered system; licensed banked restricted banks, et cetera. How does that relate to the Internet? Let's compare the two, because I see a lot of parallels between the stresses in the banking sector. We have so many banks, we have 7 million people, not that many people, who actually are we banking for? We are actually banking for the rest of the region. So are there comparisons between the stress on banks, some of them are closing their branch offices, you have probably experienced that back home. It takes a lot of money to keep a branch open, keep it staffed, so that people can get access to their money, especially in rural areas. Do you have compliance? You certainly do; as ISP, you have compliance. Perhaps even more heavily 9 regulated than ISPs is the financial sector, the banks themselves. I view these businesses as basically a bookkeeping service on top of a network. What would you do on the network, whether or not you sell content is interesting. Are there settlements between banks, yes, settlement to ISPs, yes. Do you know need to know your customer ... space? Absolutely. Certainly for law enforcement purposes. The question that I have is: what is the finesse of knowing your customer online? For banks, they have a very strict requirement for knowing who the online identity is; why, for anti money laundering purposes and for law enforcement. To me, online identity for ISPs is perhaps less clear. The tragedy here as I see it or the writing on the wall is I think there is some stress in terms of the diversity of ISPs and diversity of banks, maybe we have too many banks. But for ISPs 10 themselves, how do we measure the health of this whole industry over the last 20 years we have been liberalizing? Some may view that the payment infrastructure is really a utility. What I mean by utility is that it's low cost, it's reliable, predictable. There are low margins, because it needs to happen, society needs to function. The low barriers of entry to builded a value on top of it and to innovate, so-called as we do on the Internet, permission as innovation. Just like the electricity system. Why? Because there's a much wider societal benefit. The downside to that, unfortunately, if you are a utility and, as Geoff has told us for the last 15 years, is you have to scale. Unfortunately very low profit. You have to get big real quick or you have to find your niche to add value. The question is and the banking sector in 2007, 2008, we had two big to fail? Do you remember that? Those banks are even bigger. Part of the problem 11 there was because some of the payments stuff, the utility retail banking side, just like your points of presence, is really low margin. In fact, it cost them money to operate that stuff. So they have to go so-called in the investment side to generate that income. The customer choice, that's an open question and it's all been about, I would hope, your value in the last mile. Who owns that customer? I believe right now the writing is on the wall for the ISP industry. Because I think the business model is changing. I think everyone in this room probably understands the significance of the Comcast/Netflix deal. Is that correct? Right. This is in some sense a very subtle writing on the wall, where to get access to so-called my customers, sorry, I'm not on v6 or in fact Comcast is one of the largest v6 deployments in the states which is interesting. But to get that quality to the eyeballs for Netflix, 12 there has been this deals that going to be very interesting. Comcast is very interesting. Great network, they are doing the right thing, v6 deployment. Is that going to be the model for the future where the larger ISPs have we basically over the last 20 years we have become daddy who we have been fighting against. Are we really now the new telcos? But have we gone up a layer? From basically the message ever, the transport layer to content? That's very interesting to me. I think it's going to be very interesting to see how that plays out, both in the regulatory position and from the survival of those ISPs who serve a niche market or who are not national. They are going to have to try and find a value proposition. What I would like to explore here is the very value proposition for those ISPs to be in the business of identity and to be part of a new global payment network, which we are trying to build in 13 a community group inside the W3C. Why is this important? I think the diversity is important. I think the ability to have low barriers of entry so that people can innovate, the so-called permissionless innovation. If I'm a content provider going on the Internet, my customers pay to get on the Internet. Are we moving, in Leo's words, to a model in which you not only have to pay to get on the Internet, but also have to pay to get off the Internet? The person who is receiving that data has to pay twice. It's unclear to me. The health of the ISP ecosystem is something that -- I happened to have breakfast with Leo from ICANN, the IANA side, and we talked about this morning -- and this is kind of a lead-in, I hope, for Geoff or perhaps Geoff you can take this up -- how are you going to measure ISP ecosystem health? One thing we were brainstorming this morning is the number of people per ASN. So we looked at Netherlands, 14 population 16.8, number of ASNs 734; china, et cetera,. Interestingly, Hong Kong we have 15,000 approximately people per ASN. That's a lot of diversity and, as you probably will see in tomorrow's APNIC meeting, you will have a lot of new members of APNIC who are also from Hong Kong. Maybe there's a link there with some of the things that we are doing. The main point here is how do we preserve permissionless innovation when you as a business have to go up and scale and get large and how do you keep that ecosystem where if you're in the business of content distribution or if you're in the business of co-location where the content provider is buying a cage from you or a rack from you, to reach everyone, when perhaps -- and I'm speculating completely here -- the larger access networks later on might say, hey, sorry, you are going to have to house your equipment on my network or minimally 15 pay access to my customers. Here, the parallel I wish to draw is some of the crunch time in ISP land, I think it's also happening in finance land, in the area of financial technology. I'll go through here. I think this provides an opportunity for us in this part of the world. I would like to begin here with a question, which is in places like Africa where they don't have a banking infrastructure, mobile phone companies serve the purpose of a payment network. So my question, my operating question is: are ISPs the new banks with some new technology that we're developing? Or, alternatively, are banks the new ISPs? This goes back to your exit model as an ISP. If you can't go up and get large, who are you going to sell to? If you want to exit that way. Why? Because again there's some similarities of how banks operate, how they need to know your customer and you ISPs know your customer. 16 Are there similarities there? I think so. Interestingly, we have other payment developments outside the traditional banking network, large payment networks from people who sell stuff. In China we have Alipay, part of Alibaba; in the United States you have Amazon, with Amazon payments. These web companies, these Internet companies, are now doing payment functions. Why? Because they have critical mass. How do the banks view that? Are they under threat? Here, what I would like to look at, this is recent history last year and I find this quite fascinating because the US Federal Reserve had a call, an open call, for looking at how to improve the payment system in the US. Why? Because their payment infrastructure was not necessarily meeting their needs. The expectation these days by the youth is I want to pay and it has to happen in milliseconds. 17 Weeks isn't going to cut it. There is a user demand there. Not only that -- and the Federal Reserve is an example of a central bank -- there is the mother central bank, this hierarchy of central banks and at the top of it is the central bank of central banks, the Bank of International Settlements, and they too issued a report last year on payments and settlement systems. The banking system, the banking group here is actively trying to innovate, is actively trying to change because customer needs are changing. They're innovating, but there is another space outside of this, so-called the bitcoin world, the cryptographic world, the world which is trying to generate new forms of value which I find incredibly interesting; mathematically absolutely, but in terms of a new way of generating value. But as we have seen and as I asked earlier, there are problems. You can 18 open the newspaper and read that there are people claiming, "Look, my money is still with Mt Gox," this bitcoin exchange in Japan. So we are making mistakes again, just like we did in the ISP industry in the mid-1990s. Here if we look at them, in terms of how the technology in the financial technology space develops, there are challenges and opportunities. In the old world we have both similar, they have to your customer. You as ISPs know who you connecting to your networks, the banks have to know who they connect to their network. The status quo is the existing system, they have the global network SWIFT, but they have 1970s technology -- automated clearing houses, credit cards. To the developed world, that's fine, but to the developing world, they may not have access to credit cards and it probably perhaps doesn't make a difference. In areas where there are established banking, they have lots of rules and 19 regulations, ie really heavily regulated. When this disruptive technology comes along with new value generation, such as bitcoin, what are the regulatory problems? We can see that now in Japan, where the regulators say, "Look, the bitcoin world is not regulated, it has nothing to do with us." Yet, in another country, you will need a money transmittal licence and all that rigmarole to jump through in order to operate this kind of business. Just like in the 1990s when we started the ISP world, "I'm sorry, if you wanted to do this internet thing, you have to become a carrier." Right? No, I'm not going to do that. We are just going to get on with that and have the evolution and policy evolve. Within the new entrants in terms of this technology development and financial technology, there are new forms of value. There is this cryptographic country, there is opportunity for new status quo policy but it is going to be messy . 20 The traditional way within the financial technology sector for innovating is driven by the banks and financial institutions naturally. Their business model is to take a transaction cost, capture the money, and there is incremental innovation at the core of this huge network that they have to upgrade, and that is difficult, as we know. It's extremely expensive, given legacy systems, and it is slow. They have very highly regulated with know your customer, whereas Internet companies, technology companies, per se, they are not banks. They innovate at the edge. All this innovation with PayPal, et cetera, all this great stuff is happening because the network doesn't really care. It's inexpensive to do. There's no legacy to protect. The innovation cycle is fast. But similarly, we have very clear know your customer rules. Last year, in terms of the innovation 21 in the finance technology space, where there is no incumbent, there is no banking system that has a vested interest in its preservation -- it does, but perhaps very small part of the economy. With these new technologies, great innovation in places like Kenya, mobile payment MPeso. The opportunity here, at least for some of us in this worldwide web consortium web payments community group, is to basically make the web the new payment network; make it as easy to make global payment as it is to send email; and to drive the cost of making that transaction down to zero. Some of us believe that, if you do that, we are going to have a new boom, new business models. For example, if you're a blogger and a lot of people subscribe to your blog and if you charge 10 cents and you have 100,000 customers, you can survive. Similarly, as we experienced in Hong Kong, trying to do new forms of trade with intangibles, with intellectual property, the long tail 22 side, the price point matters, but the cost of routing that money is currently very difficult. The W3C I don't need to go through. The web payment community group we have a website. The approach that we are doing is very different. Right now, you have these silos, these payment and software marketplaces that are particular to the platform, either IOS or Android. We want to try and reinvigorate the web in many senses, by having the programmers who program for the web have automatic access, so that on a single device, your wallet doesn't follow the device. It can be somewhere else. In fact, in your wallet, you will choose the payment mechanism that you wish to do. The interesting thing is that we built on mozPay, Mozilla. This basic technology is already in the Firefox phone. The tech is done, the policy side is not done. This morning you spent the whole morning talking about policy. What I'm trying to allude to here is 23 that I don't think the innovation on the Internet is done by a long shot, so-called Internet governance. This is a great example of a new space where there needs to be some coordination and policy at the global layer, because in some sense the technology is already there. If you can just bring up the web payments website, please. The URL is webpayments.org. This is payment technologies on the web. What we are trying to do is build payments into the core infrastructure, just like for HTML. If we can put payment inside the core infrastructure for the web, that's what this group of 100-odd people are trying to do. There is code there. We want to build a better web. The approach is the same as deployment. What we have basically got are six different protocols. The most interesting thing and what we are trying to do right now is you have 24 this closed world where there are proprietary standards for payment and you have people in the open world who are doing open standards. What we are trying to do is bring those two worlds together, in a very simple way, to have a very standard form for request of payment and a very standard form for digital receipts. Where are we? This is very early days. If we can bring up the next slide, which is the workshop tab. Right now, this is a community group which just has no standing, we have a workshop next month in Paris, we know we are not supposed to do sales here in APRICOT -- it is not, because this event is already full. This event is trying to bring together the SWIFT group, people who use traditional payments, banks, regulators, we have the US Federal Reserve coming. We are trying to resolve the policy issues on how to do this very simple thing; to have a very common way of initiating the request and having 25 digital receipts. The interesting thing with the technology that's been developed is really two things in my view. The first is that if you list your product or service now on the Internet, it's no use if it can't be discovered. A lot of the RDFA and RDF technologies are already built in there. The other interesting thing is that with respect to your digital receipts and all your financial information, that follows you. It's not your bank. It's not your -- you have control of your identity information. What that may mean in future is that if you wish to move banks, you can just take all your information with you, but your financial history and move to another bank, because that information is yours. That's why this is slightly different from the traditional models, because the role of the identity is different. Where this is interesting to ISPs in terms of your exit strategy is, again, 26 since just like the banks know your customer, perhaps the role of an ISP in future is no longer about bandwidth or Internet addresses, it's about managing the very subtle interface between maintaining the identity of the individual or the business in meet space ^ and that online. As we know, online we can have multiple identities. The roadmap here is actually this group has been existing since the financial crisis, 2008. We help with defining RDFA, the resource distribution framework attributes aspect. In 2009, we started work on JSON LD. I think everyone knows that. That was one of the outcomes, in terms of link data. It's very heavily used in public sector information. People who have joined the group are those including people who are developing the browser, the browser manufacturers, people in the financial sector like Bloomberg and people in the new area of cryptographic currencies. 27 The interesting thing here is that we have presented last year at the Global Banking Conference for SWIFT, that I talked about earlier, in Dubai and what we are trying to do is bring together the two communities who innovate in finance with the traditional web communities. As I said, last year we had these series of events in Africa and Hong Kong, et cetera. In March next month we are going to have this first working group. Hopefully we'll hash out again the details of the spec. It's all online, you can download the code which is also there, so that we can resolve exactly what the needs are. The view here is, after that, to formally form the working group -- I think we have critical mass to do so; finalize the specs within 2016; and release it as the first world standard in 2017. That's the roadmap here. The opportunity as I see it for our part of the world is to get in at the ground 28 floor. As part of our working group, we have I think the second largest payment provider from Korea. CNNIC is participating in this Conference in Paris, and several other people from the world. This is an early emergent area and the opportunity is we can start to look at this area, look at what happened, for example, in places like Hong Kong, where our infrastructure is built out, we are done as far as infrastructure. We are going to have to look at the content layer, which everyone is looking at now, how to get value out of that content layer. It may be movies, but if you want to pay for movies, is that going to be inside your network or can we do something on the web? The key thing here is the rate of change. Traditional financial networks have a certain rate of change. We believe that the people in the working group can, and the tech companies, the members of the W3C can innovate, perhaps at a faster 29 rate. With that, thank you very much. I'll take questions at this point, if any. If not, I'll hand over to Geoff. Thank you APPLAUSE >>Krystal Waine: Thank you, Pindar. Philip Smith, could you please come to the stage and present Pindar with just a small token of our appreciation. APPLAUSE Thank you very much for the enlightening talk APPLAUSE Our next speaker is Geoff Huston, Chief Scientist from APNIC, and he will be giving us an update on the BGP in 2013 including the growth of the eBGP routing table over time and some projections of future growth. >>Geoff Huston: Good afternoon, all. This is me. I'm Geoff Huston, I'm with APNIC. You know, Pindar, I couldn't help thinking of John Law -- 1691 to 1729, I believe -- who wandered around Europe 30 basically saying this whole thing about silver and gold, you know, we could just exchange bits of paper as promissory notes as a promise of silver and gold. You don't need to move the money. Revolutionary. Unfortunately, his fate was also somewhat revolutionary; a man who was appreciated by history and not in his own time. Pindar, I think your project has much better chance of success. But it is a period of time, and paper no longer cuts it. Indeed, I was reflecting on that and thinking, you know, it is a message for our time. But from the prosaic down to the technical, because this next talk is entirely different and, unlike Pindar's excellent summary and synthesis of ideas, this one is down, dirty and detailed. However, hidden in here is a gem that I think is going to get me the Nobel Prize for Physics, so we will see how we go. The reason I'm talking here is about 31 the APNIC policy process. We distribute addresses. But the reason we think so hard about it is to make sure that other parts of the Internet work, and a part of that, a major part, is the routing system. Because if we don't have a system that can route these addresses, the addresses have no point. We talk about aggregation. We talk about the idea that the addresses should follow the topology of the network. If the network isn't organised along nation state lines, neither should address distribution. We talk about address policy that try and say addresses need to be where things that require addresses should be, because that's the way we make the routing system work. But together with that quite pragmatic view, we also carry a huge load of mythology. Some of that mythology is kind of weird and we work that if we repeat it often enough to ourselves, if we colour the paper with enough colours, 32 it attains a status of money. Just a mantra. You can burn it if you want, it makes no difference. In routing, we believe the routing system is a microsecond away from destruction. This research paper talks about the rapid and sustained growth of the Internet. It talks about all these risks and all these nasty little de-aggregators and the routing system is going to die in any second's time. You kind of think, we say it to ourselves, but how true is it? This is actually an exploration of trying to undercut those kinds of common mantras and myths with some real data. To actually look at this and say, what can we see inside the routing system and are we living on the edge of stability or quite comfortable? This is my version and this isn't SWIFT, this really is the Internet, up and to the right, starts at 0 and currently sitting at half a million -- we win. 33 You can actually see, too, the global financial crisis is hidden in there around 2008. The Internet is just business. Things that affected business are visible too. We actually caused our own boom and bust -- go, us -- in 2000 and 2001, and that's in the routing table. So a lot of these things that happen in the wider world are equally very visible inside the world of BGP. That's certainly what we see, even in the recent past. There we see the count over the last couple of years and, interestingly, this is after we ran out in Asia Pacific. Oddly enough, exhaustion hasn't dimmed your optimism. You're routing ever finer quantities, but the routing table still grows. Occasionally, people just have brainstorms and go, "err". What we noticed inside our particular observation system is that on one day, 25 June, all of a sudden a whole bunch of routes just appeared. Oddly enough, it was only just 34 us, and I blamed Martin Levy over at Hurricane Electric because he certainly gave me all these routes, and he immediately blamed AS 7029, which happens to be Windstream, for giving him the bulk of these routes. Both true. What it really says is that in the routing table there is no God. There is no single view that trumps everyone else. Everyone's view is relative and this is a view from me. The slopes and trends are kind of interesting and certainly last year was fits and bursts. We tapered off in April and May and then started doing stuff again in July, tapered off in October and then at the end of the year another burst. When I look at other folk, I certainly see I had an aberration in the middle of the year, but generally slow and steady growth is pretty typical across the entire Internet. Where do those addresses go? We ran out. What's going on? I can look at the 35 amount of addresses routed through the year and there I see evidence that we really did run out. But what's going on in that all of us got surprised about? We thought -- or at least I thought, I can't speak for anyone else -- that as we ran out of addresses and allowed folk to monetize addresses and use those back into the market, that this would fuel the continued growth; that we seem to be needing a lot of addresses and that somehow those unadvertised addresses would get used. That purple line at the bottom is the number of unadvertised addresses and, interestingly, it's been stable since 2011. Whatever price the market has achieved has not been the price to flush them out. Buyers and sellers aren't making the matches at the level of volume that would radically change that picture. What's going on now is that the unadvertised pool stays steady at around 50 /8s, that's 800 million addresses, and 36 nothing much is happening. One-fifth of the space remains locked up. In other news that's going to surprise and shock you, ARIN and LACNIC are going to run out this year. But we all knew that. The routed address span over the last year, people are mucking around with /8s, but in the last year or so, we continue to see more addresses hitting the routed network. The number of ASs, slow, steady growth. Put all those together and what we see is the average prefix size is getting smaller. This whole idea that we needed to aggregate in routing isn't true any more. The amount of router prefix is getting finer and finer. Those folk who are thinking that maybe, instead of routing addresses, you can route unstructured names or route money, you're not far away. Because the capabilities of this system are truly awesome. What we're truly doing is routing finer addresses and accommodating 37 in the system that seems to scale brilliantly. Overall metrics for last year: the Internet isn't growing by 100 per cent a year any more, it isn't even growing by 30 per cent, the basic metric is about 8 to 10 per cent a year at this point. Population grows at 3 per cent. The Internet grows at 8 per cent. Life is normal. We are nothing special any more. We are just business. That's v4. V6, wow, all that proselytizing, all that v6 is wonderful. There is World IPv6 Day. There is the long-term results. Geez. There was a slow and steady growth that World IPv6 Day didn't actually alter. V6 has its own trajectory. It's slow and deliberate. It is not fantastically high. There's a slight kick at the edge of the year, but in terms of the number of routes I see, 16,000-odd routes. Even in terms of address space, it's 38 slow and steady growth, which is not what we saw earlier. All of those metrics -- AS count, that's not good. That's tapering off. That's meant to be up, not down, guys -- up, more. What's the growth schedules? Certainly better than v4, growing at a rate of 20 to 40 per cent per year. But the bad news is that at this rate, for ASs, we're finally going to get universal deployment in 2030. That's not possible. Not going to happen. It's just too long. Something needs to change inside all of that. We need to make that change. Many of you run routers, many of you buy them from routing vendors. Most of you seem to think that your routers are going to last for some years, not months. How big a router do you buy? What should you be looking at? Despite the fact we ran out of addresses, we can probably make some projections here. This is the first order differential. I'm using the power 39 of maths. Oddly enough, the first order differential is not growing fantastically. Most of the models for v4 come out best at a model that looks strongly linear. When you try and put other models on it, like exponential growth or similar, the fit just doesn't seem to be very convincing; it seems to grow too fast; and, quite frankly, the data doesn't support it. At this point, I would say, with a pretty high confidence interval, that we are growing at around 50,000 new entries per year and that's just grinding its way through the system and so the projections for the next few years see us slowly growing from around half a million entries to 750,000 entries in the table. It's possibly in an exponential model, it could be as high as a million. I don't describe that as a very high level of confidence. That's an outlier prediction. V6, different kind of trajectory, more noise, more spurts. More noise, but 40 certainly growing a little faster. Those are the projections: the exponential, the linear, the quadratic. Quite frankly, the exponential still looks credible and we might get over the same period up to about 127,000 entries. Why does this matter? Because it's the cost of routing that matters. This is all about Moore's law because, quite frankly, if you are growing faster than Moore's law, you are growing faster than silicon at constant cost. All of a sudden you have to pour money in; the unit cost of routing gets higher, not lower, and that means paying for everybody. Moore's law is kind of a useful threshold here. Over the last few years, since 1971, technology has largely tracked this. This is doubling between every 18 months to 24 months and we can apply that curve to BGP. There is the application of the curve of BGP to Moore's law -- no cause for concern. So it's certainly not saying that Moore's 41 law is an issue here. V6, we are kind of growing at an exponential level something approaching Moore's law, that's certainly true, but the numbers are tiny. If you are going to hit the big red alarm bell, you are not going to hit it over this. Quite frankly, the current routing system is growing with absolutely no pain. This is not the reason why we would press the alarm bell. But some folk argue it's not the size, it's what you do with it that counts. So what we are looking at is what do we do with BGP? How much does it have instability? Any distance vector protocol grows badly. Distance vector protocols are as noisy as hell and the noise is actually relative to the population. In theory, what we should see is up and to the right. The number of announcements and withdrawals per day should be hitting the ceiling. That's flat. This is not right. Either I'm 42 wrong, you're wrong or the Internet is doing something completely surprising. Because since 2008, 100,000 updates a day. Everyone sees it all over the planet. This is very, very strange. Who is doing updates? Who are the naughty people? Who are the naughty boys of the Internet? Because only a few of you cause noise on any day. That room has 20,000 people in it, not 50,000, not 100,000, but 20,000. They are self-selecting naughty people. The growth of the noise level is much, much smaller than the growth of the routing table. It's the log value of the routing table. That is getting bizarre. That in inside a constantly growing pattern, the underlying characteristics of routing are almost flat. Even more bizarre is the performance of BGP. You ever noticed when you put more cars on the road that your trip to work takes longer? You just can't drive at the same speed when there is someone 43 in front of you but you can try, but ultimately the crash is inevitable. If you whack on 500,000 more cars into this system, you should find the network gets really, really crap. It doesn't. The average time to converge in BGP has been 70 seconds since 2008. Again, you look at this and you expect damage and what you see is constancy. The technology is over-performing and it's introducing stability where there was none. This is entropy decreasing without energy. What I said about Nobel Prizes, we are getting to a new law here of cosmological constants, that somewhere inside this system is an element of stability that is entirely unpredicted and the reasons are not totally obvious. But we can find one or two, if you look very, very hard. If you're a new ISP, who do you connect to? The obvious answer is I try and connect as close as possible to everybody else. The network has never become longer and stringier. 44 As we cram more and more people into the network, it's diameter has actually been constant. Folk tend to drive their connections into the centre. The long haul submarine cables have more and more tenants on them. Folk build their own networks to interconnect more and more richly and the result is that the average AS path link ? to span the Internet remains largely constant. This is a graph of all of the peers of Route Views every hour since 1996. There's a lot of data behind those colours. But what it really says is that the physics of the Internet are different from the physics of telephony. The physics of the Internet having managed to breed this extraordinary rich interconnectivity and as we increase that interconnectivity, oddly enough, BGP, the routing protocol, works better. It manages to cope with an ever larger pool of people but perform convergence in 70 seconds irrespective. That's fantastic. 45 Inside all of that, I really can't be concerned. V4 has been pummelling away, doing the internet and routing it without any particular cause for concern. What about v6? It was growing. From 2008 to 2011, apart from when my router lost it, it just simply grew. But after 2011, it starts to look like v4. Maybe v6 truly has come of age, that all of a sudden the underlying metrics of growth are constant. Is there a bad boy room in v6? Yes, there is. How many folk are in that room? Less than 1,000 prefixes every day. Which 1,000? I don't know. You don't know. But that's all it is. There's this curious mode that instability is a limited function, not a general condition. How long does it take to converge in v6? Slightly longer, 90 seconds. That's nothing. What's going on is the same topology, the same average AS path length is visible in v6 and v4. There was always this theory that 46 when the topology of v6 mapped the topology of v4, v6 was following the money. As soon as v6 is following the money, it's not a toy any more. It's what people exchange, peer and settle on. All of a sudden now we can actually claim that, lookinglying at that figure and looking at the topologies ofV4 and v6, it's not a toy, you don't give it away. It's part of the way in which interconnecttivity business is traded, exchanged and bartered, exactly the same as v4, and the outcome is the same. You want to get close to the middle and, as you get close to the middle, the overall performance of the system stays constant and doesn't get worse. That just says all that. That's terrific. The next thing is all about conservatism and behaviour. How good are we at making sure only aggregates get routed? To what extent do we still say that de-aggregating is polluting the internet, causing garbage for everyone. 47 Who is advertising more specifics and is that good or bad? Here is Relcom. I didn't call them out for anything in particular, but there they are advertising a bunch of /24 because they can. Here is the folk who do it. Windstream is on this list. Telekomunikasi Indonesia was No. 3 on that day, and so on and so forth, Korea Telecom. These folk all advertise more specifics, and there are a lot of them. These are unnecessary in the grand scheme of things. They don't add information to the routing system. They simply make finer grained objects and the real question is: is this noise or value? When you look at this, you look at the number of more specifics in the routing table, since 2002. It looks much the same. Let's check that theory. This is percentages. Since 2002, 50 per cent of the Internet is more specifics. That's flat. How do we get there? Interesting, 48 evcerybody sees this. I look at Route Views, that's flat as well. How much address space? It increases, but who is doing it? Only 332 origin ASs do 50 per cent of these more specifics. It's a very small bunch of people advertising half of the routing slots and the address space. If this was bitcoins, they would be as rich as hell, because slots make money. These 332 are doing an extraordinary amount of more specifics. It's a very skewed distribution. The top 20 ASs, 40,000 of them. Here is much the same list, you have seen it n before. In v6, what's going on? Like I said, v6 has come of age, more specifics are now in v6 as well. It's not quite 50 per cent, we are up at 30 per cent and growing. We will get there sooner or later. Does everyone see that in v6? Yes, Route View says everyone sees it. How much address space? Well, low at the moment, but it's growing, up and to 49 the right. Are we getting better or worse? The message that we keep on pummelling in the routing community is, don't pollute the planet? This is just crap. We have all got to take it. If you send out more specifics, you are damaging the entire world. So I took the top 10 ASs, every single day, who had advertised more specifics and I tracked their history for three years. This is the list of folk who are in that list -- not that many. If you review the slides and see yourself there, I'm watching you. I look at your track record over three years. Some folk are up and to the right. Some folk are down and to the right. Some folk are steady. Some folk do step functions. That's Brownie in motion, isn't it? That's just random. It's just individual behaviour. So we're not getting any better, we are not getting any dumber. 50 But out of that random noise comes a flat 50 per cent. So again, out of complete random noise, with no energy, entropy is decreasing. The order in the universe is getting better. This is one of these theories of the Internet that, cosmologically speaking, there is no heat death in support for us. Without energy, we are increasing the order in the Internet rather than its disorder. This is bizarre and Nobel Prize winning, I am sure. At some point, what's going on is, aside all of this random noise, it's not getting any better but it's not getting any worse. There is this sort of evenness about this entire system that is simply strange. But the issue is there are these smaller number of people doing all these specifics. Are they noisy people? Are they the people who are those 20,000 noisy prefixes? Where do I find updates? So I can look at this and actually categorize these more specifics into red, 51 blue and green. Red is senseless routing vandalism that has no purpose in being there. You know who you are. Blue is hole punching, that I have taken a bunch of addresses from my upstream and I'm advertising in a different way, and the green is traffic engineering. Those are the types of more specifics and, in general, senseless routing vandalism is about 40 per cent of the problem, hole punching is about 30 per cent, traffic engineering is about 25. More specifics are traffic engineering? Not really. Just a very small amount. Who generate the updates -- the vandals, the hole punchers or the engineers? Let's find out. Because we can look at the daily updates and go, "Wow, this is interesting." Even though more specifics are 50 per cent of the routing table, they're 80 per cent of the updates. So, yes, these more specifics are the culprits. 52 They make up the bulk of what's going on inside routing. Those small number of ASs are indeed responsible for most of the noise. What sort of specifics are doing this? I can do a few sort of analyses of these and find that it's actually not the vandals who are the worst at this, it's the hole punchers. I don't know why, it's not obvious to me what's going on, but, yes, those more specifics do generate more noise and the hole punchers generate more noise. I can do the stability metric and show you much the same thing. The traffic engineers generate as much updates as I would expect. The sensing routing vandals are generally stable and my suspicion is they're advertising those specifics without any particular reason. So they are not mucking around with them. They don't change them. They're stable because they're unnecessary. The ones that actually do change a fair deal are the ones that are 53 punching holes in other people's aggregates and perhaps that's why they actually need to change things to make it work. Profile of BGP. It really is true that more specifics are our problem. They general rate around four times the updates. Of those, it's actually not the senseless routing vandals that are most to blame. It's the hole punchers that give us the most problems at EGP. Should we do something about it? Should we take those people -- some of you are here -- take them quietly out there and do something with their advertisements? Because certainly we make BGP the collective garbage dump of the Internet. We ignore it, we leave it go. Could we do better? Of course we could. We could do a lot better, because a very small number of origin ASs are responsible for a huge amount of the dynamic metrics in growth in routing and we can do a lot better. We can rationalise this and apply 54 enormous amounts of pressure on them and it will actually make BGP an even better place. We could do better. But should we? What's the problem? Is routing blowing up? No. Are your routers going to melt tomorrow? No. Is the modest amount of attention you put in BGP producing fantastic results already? Yes. So in some ways, if you said to me, "I could spend three man years of effort and get rid of my more specific, should I?, I would go, "Geez, that is a really hard question for me to answer. I can't tell." Because in some ways, even though you are doing damage and on a real scale, that's a terrible thing, don't damage the rest of the world; but, honestly, the damage isn't that slight and the equipment we run is readily able to cope with what you are doing. Don't make it any worse, keep that level of entropy constant, we are okay. Make it worse and things go bad. What it appears right now is we are 55 doing something phenomenally good in routing. It's unconscious. It's a collective action that's actually got a beneficial outcome. There are no routing police. You don't need a ticket to go into the bad boys club or the good boys club or anything else. The collective random actions of a whole bunch of folk actually produce an outcome that has an astonishing amount of stability and beneficial outcome to all of us. As I said, this is the laws of physics turned on its head. It should never have happened. It is. We don't understand why. The problem is that because we don't understand why, when we damage it, we might only realize when it's too late. Whatever you are doing, don't change. Thank you. APPLAUSE I will try and take some questions if there are any. As I said, it was a technical talk, very quickly delivered, 56 for which I apologize >>Krystal Waine: Thank you, Geoff. I would like to now invite the Chair of APNIC, Maemura Akinori, to give a closing note. Thank you. >>Maemura Akinori: Thank you very much. Akinori Maemura, Chair of the Executive Council of APNIC. That's really great to have the two great presentations in our Closing Plenary and I'm a bit excited to have that and I need to calm down to make my address. As well as the two excellent presentations here, I appreciate all the speakers throughout this week, who brought excellent knowledge which is indispensable for keeping the Internet work and evolving. We had countless fantastic sessions this week, but among them I would like to highlight and appreciate our partners who collaborated with APRICOT this time. ICANN, which launch Asia Pacific hub last year, brought ICANN update session 57 this time. We expect the collaboration with ICANN will increase in an organic way throughout the future. ISOC, this time kindly brought their net ops workshop, with a bunch of presentations which shows up-to-date operational technologies. APCERT joined APRICOT, this time to have sessions where a number of certs shared their practices to keep the Internet secure. Asia Pacific IPv6 taskforce session was the full line up of the IPv6 speeches for future of the Internet. This year, it is the very pivotal year for the governance of the Internet. Since ISO leaders published the Montevideo statement in October, we have 1 net initiative. We will have the NetMundial in Brazil, also known as the global multi-stakeholder meeting on the future of the Internet governance, which will be held in England, in April, as well as various United Nations and ITU activities. 58 In the session on Wednesday, the title was "From Governance To Cooperation: Decoding the Puzzle." It is really explaining title. The high-level panel tackled with the wider spectrum of Internet governance issues and had a really substantial discussion there. All of them are all about how we run the Internet now and throughout the future and all those who get together here, it's you, really work on cooperation, collaboration, networking; APNIC is the place for those. I would like to ask you to not forget Friday program which is the APNIC Member Meeting which is our main business. It is not limited to the APNIC members, but open for everyone and where you can have the oral update, what is APNIC is doing right now. Thank you very much all you come here and join here to do the Internet and keep the Internet running. Thank you very much. 59 APPLAUSE >>Krystal Waine: Thanks, Akinori. Now I would like Philip to come back up on stage to say a few words of thanks. >>Philip Smith: I'm going to be slightly different, I think. What I'll do is while I give the vote of thanks, rather than as I run through thanking everybody, I think Akinori, if you would come back here. What we'll do, as I mention each of our supporters, I would like you to give them a token of our appreciation. I think if we'll do it like that, it will work a little bit better, just upset the process a little bit for you. >>Krystal Waine: You are mucking up my notes. >>Philip Smith: No, it will all work out just fine. So I don't know, usually when it's my turn up here, at the end of all this, I do the pause and I think of all the work that's gone in over the last week for everybody who has been involved with APRICOT -- I should say the last two weeks. 60 As I mentioned in my opening remarks on Monday afternoon, this Conference had to be put together in four weeks and I'm not going to go through all the reasons for that again. That was quite a significant undertaking that, as I said on Monday, we don't particularly want to repeat, because of the amount of effort that was involved. The commentary that I have heard throughout the week from many of the delegates is how much they have appreciated it. All of us here who are doing all the technical support tend to be behind the scenes, we see everything that is going wrong, we see all the issues and so forth and we are working very, very hard to make sure that we sort those before the I suppose the general delegates see any of it. But many folks have said how well the Conference has gone and nobody has commented on anything, what we have seen that hasn't quite worked how we expected it. Maybe 61 we set high expectations or too high expectations or maybe, I don't know, we are just all accustomed to things not being perfect all of the time. The first thing we really need to do, we want to thank all our sponsors. As you can see from the board here, it's been a very impressive signboard that we have managed to arrange. We managed to retain a lot of the sponsors who originally committed to support us during the event construction in Bangkok. We have gained many more supporters from the local community here in Malaysia. I think displaying them on the board like this for all of you to see is just -- I really do want to recognize their generosity in really supporting this event. We normally rely so much on having a local host to pull the sponsors in and having no local host here in Malaysia and still being able to pull a lot of those in through the efforts of the APNIC Secretariat, I think has been pretty 62 amazing. If I look at some of the sponsorship we have had over the years. That's where I thought I would just upset Krystal a little bit and mess up the process, so as we go through my vote of thanks, I just want to indicate each sponsor and give them memento as we do this. True and NecTec bid for this event two or three year ago; it has been quite a lot of time. They have been in preparation for this Conference for a significantly long time as well; right up until mid-January this year, when the state of emergency in the city of Bangkok really caused them to give us the advice to consider another location. The events in Bangkok even at the moment are still very sad for their community and of course disappointing for us as the international community, because of missing out of the Conference and because of the impact it's having on the local international industry. We have several members of the Thai 63 Internet community here and we really do appreciate their participation in the Conference in Kuala Lumpur. I really do want to recognize True and NecTec for all the work that they have put in, all the preparation, the many conference calls that we have had with them over the time. I do want to emphasize how much work they put in before we had to do the move. Internet access was provided by TM. Again, trying to put connection into a hotel in probably -- I don't know how much notice we gave them, it probably was not even four weeks notice because we were spending quite a bit of time sorting out venue and so forth before we even got that. But TM came to the fore very quickly and provided us with 150 megabits of Internet access. It hasn't all been as visible as we had hoped, certainly on the apricot-b wireless network, for those of you who are the real techies amongst us. All the more reason to provide the A22 11a 5Ghz wireless for venues like this. Those 64 folks with newer devices have had great Internet experience and those with others maybe not as good as we would have liked given the profusion of other wireless access points in the hotel. We thank TM for providing the Internet access at such short notice for us. I think I'll just carry on and we can do the certificates and so forth a little bit later, so you can sit down, Akinori. I'm sorry. It's fine. Opening reception: MCMC provided the support for the opening reception on the Monday evening over in the Pyramid Hotel. The workshops, those of us who were in the workshop, instructors and the participants, Taylor's University hosted us. I think I have gone around in circles too much today. I think Taylor's University is over that way, if you look, approximately anyway. They are very close here, 20 minutes walk, probably 20 minutes by car in traffic as well. But they are very close. We had a lovely 65 time there last week, superb host and we are very appreciative for the facility and all the effort that they made to look after us. We had Equinix doing the peering social. Equinix has been a long-term partner. We appreciate their support and we hope they carry on supporting the peering activities with APRICOT for years to come. We have many community sponsors, folks who have given all kinds of donations, big and small and helped us in various shapes and forms. We'll come back to them in a bit more detail as we hand out the tokens. Bronze sponsors, again, who have helped us with covering various parts of the Conference, the small contributions. You have seen all the sponsors, of course, have the tables outside and they have certainly had good opportunity to mix and mingle with the participants. For network hardware, we knocked on the door of Cisco. We already were 66 talking with Cisco Thailand for the Bangkok iteration and, to Cisco's credit, they managed to re-direct things to Kuala Lumpur at very short notice. They got Cisco Malaysia involved and, between the two country organizations, they again fulfilled our needs with access points and switch infrastructure and so forth. So very appreciative and thanks there. Fellowships. There were several fellows here for APRICOT. The fellowships were funded by Internet Society and the Network Start-up Resource Centre. Both organizations have a mission to improve the Internet, especially in developing countries and so on. So having them support a fellowship program so actively is really appreciated. I'll use this opportunity to ask other organizations to join Internet Society and NSRC to do likewise. The more support we can get, the more folks from less advantaged parts of the 67 Internet in Asia Pacific region we can have coming to this Conference. Internet Society also supported our newcomers event. We welcome newcomers to every single APRICOT. It's a chance for them to be not intimidated by the frequent participants, but be introduced to the Conference, get to meet, hopefully now familiar faces and feel as though they are part of the Conference itself. Several local supporters, the five organizations here -- I was going to read them out, but I can't read the first one. It's probably this screen, but MyKris, MyNOG, MyREN, FibreComm, IP Server One. So very appreciative, again, of their support, coming in so quickly to help with the Conference. I had to add this slide in. APNIC, of course, is the organiser of this. APIA has asked APNIC to run the Secretariat for APRICOT for over a year now. So APNIC very generously has supported the Closing Social which we'll enjoy tonight after we finish here. 68 Of course, supporters of the APNIC Member Meeting tomorrow, our friends at CNNIC, KISA, NIXI, JPNIC and TWNIC. Finally, the AMM closing dinner tomorrow night is supported by APJII and PHCOLO. I think we have covered all the sponsors. It has been quite a long list. I'm not accustomed to going through such a long list as this. Akinori is going to be quite busy with the recognitions. But I think I'll just finish this and then we can get on. First off, again, I had to add this in. I don't want to miss out thanking the APNIC Secretariat for doing this. One of the things when APNIC took on the Secretariat for APRICOT, was that it meant organizing this Conference. It's not just an APNIC Conference. It is APRICOT of which has many component parts, of course a major part is the APNIC 37 Conference. But missing themselves out from the set of slides here doesn't do justice to all the work that Tony and his team yet again as I said on Monday, have had to do in the 69 last four weeks to put this together. It's been pretty amazing. I need to thank the APRICOT management committee, volunteers from across the community, all the way from APIA board, staff from APNIC, people who have hosted APRICOT to the past to future hosts and any other volunteers. They turn up every month and every other week, Conference calls and so forth, to help with the organization. Program committee volunteers from the industry, we had 50 members of the program committee this year. Program committee will stay on to help with the content, technical content for APNIC 38 or at least I hope they will. We have a fellowship committee. They were working back in October, November, to decide the fellowship. We had a large number of applications, as we always do, limited funds, so I'll drop my hint about more funding for the fellowship program again. The more supporting funds we can get, the more fellows we can include. 70 Technical committee and the volunteers from the local Internet technical community, do need to thank them. They were ably let by Mas. Mas is practising for APRICOT next year. I think he realized that by now. I hope he's in the room, unless he's taking the network out from various parts of the hotel that we have to leave. He along, with John Lee from Global Transit locally, plus the other volunteers, have done a lot of work to get the network up and running for us. I would like to thank all the workshop instructors, tutorial instructors, the speakers throughout the week, participants, both who have come here, participants already on their way home, all the remote participants as well. All the local volunteer staff helping this week and of course last week we have had several volunteers as you might have seen at the registration counter running around helping with different things, the 71 staff at Taylor's University, the staff at the Sunway Hotel, they have all been most obliging, most helpful. Nothing has been too challenging for any of them and it's always been smiling face in any of the assistance. It's just such a pleasure to work with such good local support like this. I hope I have covered everybody. If I have missed anybody out, I must apologize. Before I say, "See you in Fukuoka, Japan" -- and I will get into trouble for saying this -- I must single out Molly for all the work that she's done. Please will you join me in a round of applause. APPLAUSE Molly has been the project manager for this Conference and so she has been very much the lead person to make all this happen under Tony Smith's guidance. So again, thank you all for all the work that you have had to do as a project manager, project leader, the sleepless nights, the sleepy days and so on, but 72 I think you can get a rest very soon. With that, APRICOT 2015 is in Fukuoka. The 2015 team will tell you a lot more about that in a few minutes time and I will hand back to Krystal. >>Krystal Waine: So we are handing out the certificates now? >>Philip Smith: Yes. >>Krystal Waine: Are you sure? >>Philip Smith: I can change the running sheet. >>Krystal Waine: Okay. This is my Ms America chance to hand out certificates to the sponsors. So if the sponsors could please approach the stage. First NECTEC and True, do we have representatives, then TM and MCMC, Taylor's University. APPLAUSE Equinix. APPLAUSE We have Telekom Malaysia. APPLAUSE Do we have someone from Taylor's University? Going once, going twice? 73 No. Okay, next one. Anyone from Equinix? No, okay. Biglobe representative? This is fun. Yay. APPLAUSE FullRoute. APPLAUSE BTI Systems DotAsia. Extreme Broadband. JPIX. APPLAUSE Google. APPLAUSE ICANN. APPLAUSE MyIX. SoftBank. APPLAUSE Alcatel-Lucent. APPLAUSE JPNAT. APPLAUSE Netka System. 74 Network Hardware Resale, are you in the room? NTT Communications. APPLAUSE Telstra Global. APPLAUSE BTI Systems. Cisco. ISOC, Internet Society. NSRC. APPLAUSE NIXI. APPLAUSE CNNIC. APPLAUSE TWNIC. APPLAUSE JPNIC. APPLAUSE KISA. APJII. APPLAUSE PHCOLO. APPLAUSE Fibrecomm. 75 MyKris. MyNOG. APPLAUSE MyREN. APPLAUSE Our lucky last before we find out about APRICOT next year is IP Server One. No, not here. Okay. Thank you very much. APPLAUSE If we could please have Maemura-san to talk to us about APRICOT 2015 in Fukuoka in Japan. >>Maemura Akinori: My own reason to stay in front. Video first, thank you. (video played) I try to have the eight minutes version of this promotion video, but I failed. Thank you very much, again. Akinori Maemura, this time it's for the APRICOT-APAN 2015 Japan organizing committee, which is formed by the eight main organizing members for running the 76 APRICOT-APAN. Please stand up those from the organizing committee. Yes. APPLAUSE There, there, there, yes. Thank you very much. 2015 is coming to Fukuoka. This is the first time since 2005 for APRICOT in Kyoto to Japan and this is the second joint Conference between APRICOT and APAN after very successful joint meeting in 2011 in Hong Kong. In Japan, Fukuoka is regarded as the gateway city to Asia, which is in west side of Japan, in the very near to the Korean peninsula. The Fukuoka gathers a lot of visitors from all around the region and all around the world. It is very famous for the very good food, especially the very good seafood and it is not very expensive than in Tokyo. Very cheap. You can enjoy. It is as well the urban area, with the Fukuoka city has 1.5 million population. In the end of February to 77 the beginning of March, when APRICOT-APAN 2015 will be held is actually too early for cherry blossom, but we are very happy to invite you to Fukuoka in the very good season of plum, which is the flower for Fukuoka Prefecture. We have the really big shrine, whose name is Dazaifu Tenmangu. It's really famous for the plum flowers. You will be welcome by the full bloom of the plum at the time. This is the APRICOT-APAN 2015 logo. We will be -- it will be meld from February 24 to March 6 of 2015 in Fukuoka. Japanese say, "Fukuoka ni Kinshai." "Kinshai" is "please come" in Fukuokan dialect of Japanese. I would like to place the heart felt "Fukuoka ni Kinshai" to you. I really appreciate if you come to Fukuoka to join us for APRICOT-APAN 2015. Thank you very much. APPLAUSE >>Krystal Waine: That now brings us to the end of APRICOT 2014, so thank you for all coming. 78 Just a reminder that there is a closing dinner tonight and it is Sunway Pyramid, down in Mantra Indian & Asian Restaurant at 7:00pm. This may be the end of the APRICOT Conference, but remember tomorrow there is the APNIC AMM tomorrow. Everyone is welcome to attend and get an update on APNIC's activities during the past year. Thank you, ladies and gentlemen, and hopefully see you next year. APPLAUSE Live captions/realtime transcription by mediascriptz.com