The ISP Column A column on things Internet A Year in the Life June 2013 Geoff Huston On the 6th June 2012 we held the World IPv6 Launch Day. Unlike the IPv6 event of the previous year, World IPv6 Day, where the aim was to switch on IPv6 on as many major online services as possible, the 2012 program was somewhat different. This time the effort was intended to encourage service providers to switch on IPv6 and leave it on. What has happened since then? Have we switched it on and left it on? What has changed in the world of IPv6 over the past 12 months? Who’s been doing all the work? In this article I’d like to undertake a comparison of then and now snapshots of IPv6 deployment data. For this exercise I’m using the data set that we have collected at APNIC using a broad based sampling of Internet users through online-advertisements. This approach to measurement has been described at http://www.potaroo.net/ispcol/2013-05/1x1xIPv6.html. The daily snapshots of the V6 measurement can be found at http://labs.apnic.net/dists/v6dcc.html, and the breakdown of this data by economy and by provider can be found at http://labs.apnic.net/ipv6-measurement/Economies/ and http://labs.apnic.net/ipv6-measurement/AS/. Let’s look at the big number picture first. A year ago, in June 2012, we measured some 0.60% of the world’s Internet user population that was able to successfully retrieve a dual stack web object using IPv6. At the time the estimate of the total user population of the Internet was some 2.24B users, so 0.60% equates to 13.5M users who were using a working IPv6 protocol stack, and preferring to use IPv6 when given a choice of protocols by a dual stack service. What does it look like one year later? In June 2013 We see a rolling average of 1.29% of the Internet’s users who are preferring to use IPv6 when presented with a dual stack object to fetch. With a current estimate of Internet user population of an estimated 2.43B users, that figure equates to a count of 29.3 M users. In one sense a growth of 0.60% to 1.29% of the Internet sounds like very small steps, but at the same time a growth in users from 13.5M to 29.3M users is indeed a significant achievement in 12 months, and is easily doubling the extent of IPv6 use in this period. The tracking of this metric across the past 12 months is shown in Figure 1. There is some indication that there was a significant exercise in the deployment of IPv6 in June 2012 at the time of the World IPv6 Launch event, but also some evidence of shutting IPv6 down in some parts of the network in the months thereafter. There was another cycle of growth and decline in the period November 2012 to March 2013, and another period of further growth from March 2013 until the present day. http://www.potaroo.net/ispcol/2013-06/fig1.png Figure 1 – IPv6 Deployment: June 2012 – June 2013 Where did IPv6 happen? One way to look at IPv6 deployment is by looking at IPv6 deployment efforts on a country-by-country basis. Which countries were leading the IPv6 deployment effort twelve months ago? Table 1 contains the list of the top 20 countries, ordered by percentage of the Internet user population who are showing that they can use IPv6, from June 2012. 2012 Rank Economy % using IPv6 # of IPv6 Users 1 Romania 7.40% 641,389 2 France 4.03% 2,013,920 3 Luxembourg 2.59% 12,049 4 Japan 1.75% 1,766,799 5 Slovenia 1.07% 15,175 6 United States 1.01% 2,500,684 7 China 1.01% 5,209,030 8 Croatia 0.85% 22,551 9 Switzerland 0.80% 51,575 10 Lithuania 0.66% 13,845 11 Czech Republic 0.55% 39,694 12 Norway 0.51% 23,333 13 Slovakia 0.44% 19,112 14 Russian Federation 0.39% 238,576 15 Germany 0.32% 217,494 16 Hungary 0.31% 19,896 17 Portugal 0.30% 16,406 18 Netherlands 0.27% 40,870 19 Australia 0.25% 49,425 20 Taiwan 0.24% 38,843 Table 1 – IPv6 Deployment, ranked by % of national users: June 2012 That's an interesting list. There are some economies in this list that were also rapid early adopters of the internet, such as the United States, Japan, Norway and the Netherlands, and also some of the larger economies, such as the France, Japan, the United States, the Russian Federation and Germany, who are members of the G8 (Italy, the United Kingdom and Canada are the other members of the G8). Some 15 of the 20 are European economies, and neither South America or Africa are represented on this list at all. Also surprising is the top economy at the time. The efforts in Romania earlier in 2012 to provision their fixed and mobile service network with IPv6 produced an immediate effect, and by June 2012 some 7.4% of their user base was using IPv6, after commencing the public deployment of IPv6 in late April 2012. Interestingly, in percentage terms, the numbers trail off quickly, so that only 10 countries were above the global average, and by the time you get to the 20th ranked economy in this list, Taiwan, the level of IPv6 deployment is rate was some 0.24%. So the overall picture could be described as “piecemeal”, with some significant efforts in just a small number of countries to deploy IPv6 There is another way to look at this 2012 list, which is to perform the same ranking of economies by the population of IPv6 users, as shown in the following table: 2012 Rank Economy % using IPv6 # of IPv6 Users 1 China 1.01% 5,209,030 2 United States 1.01% 2,500,684 3 France 4.03% 2,013,920 4 Japan 1.75% 1,766,799 5 Romania 7.40% 641,389 6 Russian Federation 0.39% 238,576 7 Germany 0.32% 217,494 8 Indonesia 0.17% 94,543 9 Switzerland 0.80% 51,575 10 Australia 0.25% 49,425 11 United Kingdom 0.08% 41,461 12 Netherlands 0.27% 40,870 13 Czech Republic 0.55% 39,694 14 Taiwan 0.24% 38,843 15 India 0.03% 36,881 16 Ukraine 0.21% 31,933 17 Malaysia 0.18% 30,034 18 Thailand 0.15% 27,617 19 Brazil 0.03% 26,051 20 Nigeria 0.06% 25,149 Table 2 – IPv6 Deployment, ranked by IPv6 users: June 2012 Of the 13.5M IPv6 users a year ago, some 5M were located in China, and between the four economies of China, the United States, France and Japan we can account for 85% of the total estimated IPv6 users of June 2012. This observation illustrates a somewhat fragmented approach to IPv6 adoption in mid 2012, where Internet Service Providers in small number of economies had made some significant levels of progress, while in other economies the picture of IPv6 deployment ranged from experimental or highly specialised programs through to simply non-existent. There are a number of interesting entrants in this economy list, including India, Brazil and Nigeria, which point to some levels of experimentation by some service providers in the provision of IPv6 services in other economies. Hopefully this experimentation was a precursor to subsequent wider deployment programs. Was this this case? What has happened in the ensuing year? Here are the same two tables, using IPv6 use data as of June 2013, showing a comparable perspective of IPv6 deployment as it stands today. 2013 Rank Economy % using IPv6 # of IPv6 Users 1 Romania 10.84% 1,053,237 2 Switzerland 10.72% 700,777 3 Luxembourg 6.96% 32,535 4 France 5.46% 2,824,465 5 Belgium 4.17% 339,651 6 Japan 4.13% 4,137,476 7 Germany 3.24% 2,212,062 8 United States of America 2.72% 6,768,264 9 Peru 2.42% 273,370 10 Czech Republic 2.12% 157,203 11 Singapore 1.58% 54,060 12 Norway 1.21% 53,677 13 Slovenia 0.92% 13,230 14 China 0.90% 4,651,953 15 Greece 0.78% 44,572 16 Portugal 0.76% 45,408 17 Taiwan 0.72% 120,180 18 Netherlands 0.70% 109,425 19 Australia 0.69% 121,256 20 Slovakia 0.52% 21,169 Table 3 – IPv6 Deployment, ranked by % of national users: June 2013 This table clearly shows that Switzerland, Belgium, Germany, Peru, the Czech Republic and Greece have made a significant change in their level of IPv6 deployment in the last 12 months. We now see 7 of the 20 economies as being non-European economies. Eleven of these economies have IPv6 usage rates above the global average of 1.3%, an increase of 2 since 2012. We can plot these numbers onto a world map, as shown in Figure 2, using a colouring scale from 0 to 4% of each national Internet user population that is capable of using IPv6. http://www.potaroo.net/ispcol/2013-06/fig2.png Figure 2 – IPv6 Deployment, ranked by % of national users: June 2013 The following table shows the estimated IPv6 user population per economy in June 2013. 2013 Rank Economy % using IPv6 # of IPv6 Users 1 United States of America 2.72% 6,768,264 2 China 0.90% 4,651,953 3 Japan 4.13% 4,137,476 4 France 5.46% 2,824,465 5 Germany 3.24% 2,212,062 6 Romania 10.84% 1,053,237 7 Switzerland 10.72% 700,777 8 Belgium 4.17% 339,651 9 Peru 2.42% 273,370 10 Czech Republic 2.12% 157,203 11 Russian Federation 0.21% 143,677 12 United Kingdom 0.27% 135,076 13 Australia 0.69% 121,256 14 Taiwan 0.72% 120,180 15 Netherlands 0.70% 109,425 16 Canada 0.19% 55,492 17 Singapore 1.58% 54,060 18 Norway 1.21% 53,677 19 Portugal 0.76% 45,408 20 Greece 0.78% 44,572 Table 4 – IPv6 Deployment, ranked by IPv6 users: June 2013 Again the distribution of IPv6 users appears to be somewhat is skewed, in so far as just 5 economies account for 85% of the total population of IPv6 users in June 2013, which is the same four economies of United States, Japan, France, and China, but this time joined by Germany. Unfortunately we no longer see India, Brazil or Nigeria in this top 20 economy list. The top 20 economies cut off has risen from 38,000 IPv6 users per economy to 44,000, so unless there was some form of continued expansion of IPv6 deployment (such as the United Kingdom’s rise from 41,000 users in mid-2012 to 135,000 in mid-2013), the economies at the lower end of the top 20 in 2012 were likely to slip off the top 20 list if there was no continued expansion of their IPv6 program through the year. In percentage terms what has changed over the past 12 months? The following table compares the values from mid 2012 to the values in mid-2013. The first table lists the top 20 economies who have lifted the percentage of their users who are capable of using IPv6, ranked by the rate of change of this percentage value. 2012 to 2013 Rank Economy Diff (%) Diff IPv6 User Count 1 Switzerland +9.92% +649,202 2 Luxembourg +4.37% +20,486 3 Belgium +4.07% +331,153 4 Romania +3.44% +411,848 5 Germany +2.92% +1,994,568 6 Peru +2.41% +272,327 7 Japan +2.38% +2,370,677 8 United States +1.71% +4,267,580 9 Czech Republic +1.57% +117,509 10 Singapore +1.43% +48,524 11 France +1.43% +810,545 12 Greece +0.70% +40,530 13 Norway +0.70% +30,344 14 Taiwan +0.48% +81,337 15 Portugal +0.46% +29,002 16 Australia +0.44% +71,831 17 Netherlands +0.43% +68,555 18 New Zealand +0.35% +13,174 19 South Africa +0.33% +34,022 20 Bosnia and Her. +0.32% +8,914 Table 5 – IPv6 Deployment, ranked by % of national users: change from June 2012 to June 2013 The largest change was Switzerland, where a further 10% of their users were able to use IPv6, and significant efforts were visible in Luxembourg, Belgium, Romania, Germany, Peru and Japan in terms of the ratio of IPv6 users in each economy. In terms of user population who are IPv6-capable, the table of economies who deployed IPv6 over the largest set of users is provided in Table 6. Obviously one economy where there has been substantial effort in the past 12 months has been the United States, where some additional 4.2M users are now using IPv6 in just 12 months. That is an extremely impressive effort. Similarly, there has been a significant effort in Japan and Germany. It also should be noted that as or April 2011 the further provision of IPv4 addresses through the conventional Regional Internet Registry allocation system had ceased for the Asia Pacific region, so a case could be made that the efforts this region, including those of Japan, Tawian, Australia and New Zealand were spurred on by this event. Similarly the Regional Internet Registry serving Europe and the Middle East also exhausted its pools of available IPv4 addresses in September 2012, which may have some bearing on the IPv6 efforts in Germany, France, Switzerland Romania. Belgium, the Czech Republic, the United Kingdom, the Netherlands, Greece, Norway, Portugal and Luxembourg. However, IPv4 addresses are still available for service providers in North and South America and in Africa, which makes the efforts in the United States all the more laudable for their prudence. 2012 to 2013 Rank Economy Diff (%) Diff IPv6 User Count 1 United States +1.71% +4,267,580 2 Japan +2.38% +2,370,677 3 Germany +2.92% +1,994,568 4 France +1.43% +810,545 5 Switzerland +9.92% +649,202 6 Romania +3.44% +411,848 7 Belgium +4.07% +331,153 8 Peru +2.41% +272,327 9 Czech Republic +1.57% +117,509 10 United Kingdom +0.19% +93,615 11 Taiwan +0.48% +81,337 12 Australia +0.44% +71,831 13 Netherlands +0.43% +68,555 14 Singapore +1.43% +48,524 15 Greece +0.70% +40,530 16 South Africa +0.33% +34,022 17 Canada +0.11% +33,104 18 Norway +0.70% +30,344 19 Portugal +0.46% +29,002 20 Luxembourg +4.37% +20,486 Table 6 – IPv6 Deployment, ranked by national users: change from June 2012 to June 2013 Also it would appear that Europe remains a strong focal point for IPv6 deployment at present, while the deployment in other regions is far more piecemeal. Although I must mention Peru and South Africa in this context as two highly notable exceptions to this general observation. And where is China in June 2013? What we saw in our measurements is a relative decline in the population of users who are seen to use IPv6 from June 2012 to June 2013. This decline was estimated to be some 557,000 users. One of the more variable factors for China is the role of the national firewall structure, and its capabilities with respect to IPv6, and as the IPv6 measurement system was hosted outside of China, the measurements relating to Chinese use of IPv6 are dependant on the behaviour of this filter structure. It is possible that the firewall has different behaviours for IPv6, and equally possible that these behaviours have altered over time. It could well be that an internal view of China would have a different result than that which we see from outside the country. It is also possible to provide some insights as to which ISPs are undertaking this activity, by tracing the originating Autonomous System number of the user’s IP address who have provided capability data to this measurement exercise. The following is a list of some of the larger Service Providers that are showing some significant levels of activity in the past 12 months with IPv6. The list is by no means exhaustive, but it is intended to highlight those providers that have been seen to make significant changes in their IPv6 capability measurements over the past 12 months in the economies listed in Table 6. The percentage figures provided in the list are the percentage of clients whose IP address is originated by these AS’s who are able to use IPv6 in June 2012 and in June 2013. Country AS AS Name 2012 IPv6 2013 IPv6 United States of America AS6939 Hurricane Electric 29% 37% AS22394 Cellco Partnership Verizon Wireless 6% 20% AS7018 AT&T Services 6% 15% AS3561 Savvis 1% 5% AS7922 Comcast 1% 3% Japan AS2516 KDDI 16% 27% AS18126 Chubu Telecommunications 0% 23% AS17676 Softbank 1% 4% Germany AS3320 Deutsche Telekom AG 0% 5% AS31334 Kabel Deutschland 1% 7% AS29562 Kabel BW GmbH 0% 10% France AS12322 Free SAS 19% 22% Switzerland AS67722 Swisscomm 0% 23% AS559 Switch; Swiss Education and Research 11% 18% Romania AS8708 RCS & RDS SA 11% 25% Belgium AS12392 Brutele SC 0% 33% AS2611 BELNET 3% 22% Peru AS6147 Telefonica del Peru SA 0% 3% Czech Republic AS2852 CESNET z.s.p.o. 20% 27% AS5610 Telefonica Czech Republic; a.s. 0% 4% AS51154 Internethome; s.r.o. 0% 3% United Kingdom AS786 The JNT Association (JANET() 51% 68% AS13213 UK2 Ltd 0% 23% Taiwan AS9264 Academic Sinica Network 0% 21% AS1659 Taiwan Academic Network 2% 8% Australia AS7575 Australian Academic and Research 13% 21% AS4739 Internode 5% 11% Netherlands AS3265 XS4ALL Internet BV 6% 27% Singapore AS7472 Starhub Internet Pte Ltd 0% 13% AS4773 MobileOne Ltd. 0% 10% Greece AS5408 Greek Research and Technology Network 17% 19% South Africa AS2018 TENET 0.3% 3% Canada AS6453 TATA Communications 10% 13% AS22995 Xplornet Communications Inc 0% 9% Norway AS224 Uninett; The Norwegian Research Network 16% 24% AS39832 Opera Software ASA 1% 100% AS57963 Lynet Internett 0% 56% Portugal AS3243 PT Comunicacoes S.A. 0% 1% Luxembourg AS6661 Postes et Telecommunications 4% 14% Table 7 – IPv6 Deployment 2012-2013, Selected Autonomous System Measurements What can we say about the state of IPv6 deployment one year after the commencement of the IPv6 Launch program? The encouraging news is that overall numbers of IPv6-capable end users have doubled in 12 months. The measurements presented here support an estimate that today some 30 million Internet users who will use IPv6 when they can. But this is not happening everywhere. Indeed, it is happening in a small number of countries, with still a relatively small set of service providers. What we appear to be seeing are concentrated areas of quite intense IPv6 activity. Many national academic and research networks have been highly active in supporting IPv6 deployment within their network. In the commercial networks we are seeing a number of major commercial network service operators, primarily in the United States, Japan, Germany, France, Switzerland and Romania, launch programs that integrate IPv6 services into their retail offerings. Whether this effort will provide sufficient impetus to motivate other providers to also commit to a similar program of IPv6 deployment is perhaps still an open issue today, but there is some evidence that there is now a building momentum and an emerging sense of inexorable progress with the deployment of IPv6. We’ll be continuing these measurements, and providing further insights as to where we can see IPv6 deployment underway across the Internet over the coming months. You can find daily reports of our measurements, including breakdowns by economy and tracking of progress with IPv6 for individual network service providers at http://labs.apnic.net/ipv6-measurement. If you would like to assist us in this measurement exercise, we’d obviously like to hear from you, at mailto:research@apnic.net. Disclaimer The views expressed are the authors' and not those of APNIC, unless APNIC is specifically identified as the author of the communication. APNIC will not be legally responsible in contract, tort or otherwise for any statement made in this publication. About the Author GEOFF HUSTON B.Sc., M.Sc., has been closely involved with the development of the Internet for many years, particularly within Australia, where he was responsible for the initial build of the Internet within the Australian academic and research sector. He is author of a number of Internet-related books, and has been active in the Internet Engineering Task Force for many years. www.potaroo.net