APNIC's Senior Research and Development Scientist George Michaelson gives an overview of global IPv6 trends in 2017 at the CNNIC OPM 2017 noting that IPv6 is now over the initial deployment and entering normal daily use in a significant number of economies.
3. CC IPv6 % CC IPv6 % CC IPv6 % CC IPv6 % CC IPv6 % CC IPV6 %
BE 54.40% DE 43.79% IE 43.06% US 36.20% CH 36.07% GR 34.70%
LU 32.99% EE 30.46% IN 30.27% JP 24.84% GB 24.45% PT 22.37%
FR 19.65% CA 18.53% TT 18.04% EC 17.97% MY 17.39% PE 17.32%
AU 15.58% BR 15.41% NO 14.12% NL 11.78% ZW 11.72% FI 11.66%
NZ 10.96% CZ 10.34%
CC IPv6 % CC IPv6 % CC IPv6 % CC IPv6 % CC IPv6 % CC IPv6%
RO 9.53% AT 9.12% GT 8.68% HU 8.11% SA 6.63% SI 6.32%
MO 6.27% VN 5.64% BO 5.16% SG 4.02% SE 4.01% PL 3.61%
TH 3.58% LK 3.13% BA 2.64% DK 2.42% NR 2.39% BG 2.11%
SK 1.61% IL 1.57% IT 1.43% AR 1.32% DO 1.26% KR 1.25%
RU 1.14% MX 1.09% HK 0.93% MD 0.58% UY 0.56% EG 0.43%
TW 0.42% BT 0.41% ES 0.35% CN 0.35% SC 0.33% LV 0.29%
AE 0.28% ZA 0.27% PR 0.26% IS 0.25% VU 0.21% PA 0.20%
ID 0.19% KN 0.14% AF 0.13% ZM 0.11% UA 0.10% TZ 0.10%
Above and
Below the Line:
9. Higher IPv6 capable Economies worldwide
0
20
40
60
80
100
%ofSamples
US GB DE BE EC BR IN MY VN JP
10. Top 10 by sample, US
ASN AS Name IPv6
Capable
IPv6
Preferred
# Samples
AS7922 Comcast Cable Communications, LLC 63.76% 60.38% 12719895
AS7018 ATT Services, Inc. 77.87% 71.80% 6929501
AS701 Verizon Business 0.14% 0.02% 3439816
AS20115 Charter Communications 3.07% 2.83% 3436396
AS22773 Cox Communications Inc. 42.45% 39.71% 2770177
AS209 Qwest Communications Company, LLC 0.13% 0.08% 2056507
AS20001 Time Warner Cable Internet LLC 45.61% 42.58% 2051298
AS10796 Time Warner Cable Internet LLC 32.92% 31.30% 2018570
AS32934 FACEBOOK - Facebook, Inc. 99.77% 99.35% 1802329
AS5650 Frontier Communications of America, Inc. 0.10% 0.01% 1654647
13. Lets apply this curve model to IPv6
âą Logistic Curve, a form of Sigmoid Curve
âą Use the IPv6 World adoption rate figures
âą Project forward to a (hypothetical) 100% adoptionlevel
âą Python numpy/scipy curve fit to data
âą Result?
13
16. China (.CN)
1,386,516,550 people
723,761,639 users
52% penetration
1,327 ASes
11.01T GDP
IPv4 364 in BGP
338,292,736 addresses
0.24 per head
88% visible
IPv6 69 in BGP
91,388,315 M addresses
65,912 per head
21% visible
0% capability
18. Top 10 by samples, China
ASN AS Name IPv6
Capable
IPv6
Preferred
#
Samples
AS4134 CHINANET-BACKBONE 0.25% 0.12% 6732942
AS4837 CNCGROUP China169 Backbone 0.23% 0.08% 3176306
AS4812 China Telecom (Group) 0.23% 0.07% 1357227
AS9808 Guangdong China Mobile Communication Co.Ltd. 0.08% 0.04% 1211296
AS4808 China Unicom Beijing Province Network 0.27% 0.23% 780732
AS56046 Jiangsu China Mobile communications corporation 0.20% 0.08% 506458
AS56041 Zhejiang China Mobile communications corporation 0.02% 0.00% 349418
AS56040 Guangdong China Mobile communications corp. 0.05% 0.01% 321661
AS24444 Shandong China Mobile Communication Company 0.14% 0.04% 298079
AS17621 China Unicom Shanghai network 0.20% 0.10% 183499
19. Top 10 IPv6 Capable
ASN AS Name IPv6
Capable
IPv6
Preferred
# Samples
AS21859 ZNET - Zenlayer Inc 29.35% 0.54% 184
AS7497 CSTNET-AS-AP Computer Network Information
Center
29.10% 18.18% 5281
AS4538 China Education and Research Network Center 22.64% 19.26% 71316
AS17964 Beijing Dian-Xin-Tong Network Technologies Co., 20.26% 17.05% 3149
AS4809 China Telecom Next Generation Carrier Network 4.52% 2.43% 5710
AS9929 China Netcom Corp. 2.49% 0.10% 4772
AS38019 Tianjin China Mobile Communication Company Ltd 2.09% 1.61% 17427
AS17622 China Unicom Guangzhou network 1.93% 1.55% 74198
AS4847 China Networks Inter-Exchange 1.44% 1.10% 148184
AS58466 CHINANET Guangdong province network 1.19% 0.00% 6154
20. Top 10 IPv6 Capable,Samplesize>10,000 (April-May)
ASN AS Name IPv6
Capable
IPv6
Preferred
#
Samples
AS17816 China Unicom IP network China169 Guangdong 0.88% 0.79% 146,602
China Networks Inter-Exchange 1.71% 1.37% 130,843
AS17622 China Unicom Guangzhou network 1.47% 1.25% 87,779
AS4538 China Education and Research Network Center 26.37% 22.96% 69,151
AS58466 Guangdong province network 0.54% 0.00% 35,189
AS17962 ShenZhen Topway Video Communication Co. Ltd 0.58% 0.00% 32,567
AS17623 China Unicom Shenzen network 0.61% 0.00% 29,854
AS7497 Computer Network Information Center 33.89% 11.98% 12,341
AS4809 China Telecom Next Generation Carrier Network 4.35% 2.47% 11,152
AS38019 Tianjin Mobile Communication Company Limited 10.03% 4.96% 10,534
21. Summary: IPv6 is entering its growth spurt
âą It seems clear that worldwide, IPv6 is now over the initial deployment and
entering normal daily use in a significant number of economies.
âą In our region, India, Vietnam, Japan, Australia, Singapore and New Zealand
have decided to move to a dual stack world.
âą Its possible that inside five years, half or more of the internet by end users could
be IPv6 enabled
âą APNIC continues to measure worldwide IPv6 deployment and will be reporting
on it regularly.
21
Hinweis der Redaktion
This talk is an overview of the current world trends in IPv6. Its got data up to June of this year, and is an update on the pack I presented in Beijing towards the end of last year.
The current single-line figure we have for IPv6 is quite encouraging: itâs a steady increase in IPv6 worldwide, and we are comfortably sure this is a significant viable IPv6 network.
There has been a marked up-swing in the latter part of 2016 continuing in 2017. The force of this trend is unmistakable and Iâm going to come back to this later on with a crude predictive model.
So keep this line in mind.
Now, lets look at the ranking for the leading IPv6 economies of the world, considering this line. Basically, lets work with 10% as âworld-gradeâ IPv6.
Since the last time I presented, one more economy in the Asia-Pacific region has moved into the worlds-best-practice set. So we now have 5 economies. India, Japan, Malaysia, Australia and New Zealand.
If we look in the second rank Vietnam has surged up the table, along with Singapore, Thailand and Sri Lanka. But the bottom of this list has not really altered much. Korea, Taiwan, China remain very low. There is little to no visible signs of IPv6 capability to end users.
However, if you look at the timeline, its quite encouraging how a lot of AP region economies have taken some steps. There are good signs of movement overall in our region.
Iâm going to focus on four economies: Japan, India, Vietnam and Australia.
Japan has been a consistent early adopter of IPv6, with KDDI one of the first to move. The interesting change since I last presented is AS4713, OCN which is the domestic market partner of NTT Japan. They have moved from a closed-network triple-play into a globally routed IPv6 model, and you can see they now have around 20% of their users reachable. This is a big move because OCN is one of the main providers of services to home users.
Japans overall rate of adoption is ahead of world rate, but still somewhat slow. Its not sped up significantly since launch.
Australia is still a story about only two main ISPs, Foxtel who have completed their deployment of IPv6 and are now running dual-stack, and Telstra which has moved from test phase into a fully commercial dualstack service on its cellular network, using 464XLAT. Its almost at 50% saturation, and thatâs without yet having a model which has been released to iPad and iPhone handsets so its going to increase massively once thatâs out of test phase. You can see the underlying trend for australia tracks world rate. Its perhaps speeded up a bit in its rate of growth.
Keep foxtels rate of adoption in mind:
India is now one of the worlds highest ranked IPv6 enabled economies, but its still solely due to Reliance, which has completed its deploymemnt. 80% of Reliance users are now Ipv6 enabled. The rate of adoption has a very characteristic curve the same as Foxtel, which weâre going to discuss later.
This is Vietnam. FPT looks to be the third ranked ISP nationally, and again we see this characteristic adoption curve of very rapid deployment. Its tailled off a bit lately, perhaps its to do with the logistics of deploying IPv6 in some market segment.
So if we look worldwide, we can see that other economies are doing significantly better than us. The Belgians stand out with over 50% penetration now, and the US and Germany are moving strongly at 30%. But some interesting signs are out that South American economies are also moving, including Ecuador and Brazil which now have very large deployments.
Iâve included India and Malaysia on this chart so we can see where IPv6 âtigerâ economies are going against this scale.
America is a mature market, with quite strong competition in the IPv6 segment. 5 of the top 10 by volume have significant IPv6 deployment. Iâm not sure how Facebook made it onto the list, I suspect this is a measurement glitch but its probable they now offer some kind of advert brokering service to browsers, which weâre picking up. Its interesting its dualstack and at a very very high rate of capability, which makes me think its coming from their core network.
What you can see from the US market is this regular pattern of initially slow, then quite accellerated, then again slow termination of adoption. This is a characteristic model of new deployment weâve seen worldwide. So lets talk about this a bit.
This is a chart from the New York Times, subsequently picked up by the Harvard Business Review. And it looks at the rate of adoption of new technologies across the 20th century.
So if you look at something like the telephone, it was an amazing luxury when it first came to market. A few millionaires had them, then government departments like the railroad adopted them, then slowly it moved into normal domestic life but it didnât reach 80% until the 1960s. Thatâs a long slow curve. Radio on the other hand, BAM! It came to market in the 1920s and went off like a rocket. It reached market saturation in almost 10 years. Color TVs were the same, as soon as they became affordable, they deployed in a few years. As we move to the right it looks to me like the rate of adoption sometimes is almost completely over in 5 years. The VCR for instance, or Cellphones.
So, this curve is very characteristic. Lets use it as a model, to see where Ipv6 adoption might be going.
The curve is called a logistic curve. Itâs a type of sigmoid. I used a python data modelling package, to apply it to the data we have for worldwide IPv6 capability, and itâs a very good fit for the current trends.
This is the application of a logistic function, to try and map out the future course of IPv6 adoption worldwide. Logistic functions often apply to supply-chain deployment of innovation, because they have slower initial release, then a surge of deployment, then tail off to a maximum. We have seen this historically with lots of different kinds of technology, and the pace of change seems to get faster as time goes on. The chart is HIGHLY SPECULATIVE and the actual future course is uncertain. It will almost certainly max out before 100%, and so the actual curve may reflect approach to a more plausible target like 65% or 80%
Have a look at the web if you want to see more detail about IPv6 deployment worldwide, and I especially recommend Fred Bakers research which has been edited by the internet society. Itâs a great read.
So.. Lets look at china.
The good news here, is that the volume of visible IPv6 announcement has surged from last time. When I spoke in Beijing last year, IPv6 visibility was under 10% but now its pretty respectible.
Its still a small percentage of active ASN which announce it, but much more of the national address pool is visibly being routed. Thatâs a good thing. There has been a little increase in the ASN, you have a small number more active BGP speakers, but thereâs been next to no overall increase in IPv4 address holdings at scale. There might be a lot more people with a /22 but since thatâs only 1024 addresses, the overall impact on the national address stock is small. So, the overall story remains true: you cannot have enough globally visible Ipv4 to give addresses to every device in china, now or in the future. But, in IPv6 its more than possible to do this. There is no shortage of addresses in Ipv6, and there is no impediment to a national IPv6 network from supply.
There hasnât been any visible change in visible end user access to IPv6. we understand this, its not a surprise this hasnât lifted yet, there are well understood national strategic reasons for this.
If we look at the sample counts, these are the significant ASN and entities we see in China from the outside. China Mobile is well represented in ths random sample. So, when there is a decision to deploy IPv6 we expect it to be very rapid nationwide, based on the federal/state model of ASN assignment to each sub-entity.
Ignoring the size for now, the capability to do IPv6 lies in these companies and entities. Its clear there is going to be a lot of IPv6 capable people in the countty when it goes public.
An earlier sample I took counted only the entities with more than 10,000 samples, and this is the current data I have on their IPv6 capability.