IPv6 adoption - What went wrong with IPv6?

I apologize. In February 2011 I wrote “I think now is a good time to start learning IPv6.” If you followed my advice, you probably wasted your time.

Since I wrote the IPv6 tutorial almost four years ago, I never again had to deal with IPv6. I think I forgot almost everything I learned when I wrote the guide. This article by the IPv6 evangelist Tom Coffeen reminded me that the alleged IPv4 successor is still out there.

IPv6 adoption ^

I was always a bit skeptical about IPv6, but in 2011, I somehow believed that the breakthrough of the new Internet protocol was imminent. If you regularly read 4ysops, you know that I always supported the aggressive adoption of new technologies. If someone gave me a switch that allowed me to block IPv4 traffic worldwide, I wouldn’t hesitate to push the button (maybe with a few months forewarning).

However, the Internet community obviously has little interest in IPv6. Of course, Tom Coffeen sees this different. He presents data from a few US Internet providers that appear to indicate that IPv6 adoption is growing rapidly. According to Coffeen, Verizon’s IPv6 traffic reached 54 percent, Comcast’s 30 percent, AT&T’s 21 percent, and Time Warner Cable’s 10 percent.

These numbers seem to be a bit high for me, considering that Wikipedia states that IPv4 still accounts for 96 percent of the worldwide Internet traffic.


Anyway, the Internet traffic as such tells little about the adoption of IPv6. I don’t have statistical data, but my guess is that most of the IPv6 traffic is server-to-server or router-to-router communications, because the vast majority of end user devices still run on IPv4. Naturally, servers exchange a lot of data. Hence, the sheer amount of traffic has little informative value with the regard to IPv6 adoption.

This is somewhat ironic, because the major motivation behind IPv6 once was the alleged shortage of IPv4 addresses for end user devices. About 20 years after IPv6 was developed (between 1993 and 1994), all those scenarios of the Internet’s demise never happened.

20 years is a lot in our technological world. In 1996, when the IPv6 RFCs were released, the first “mobile web” was introduced by Nokia (Nokia 9000 Communicator) in Finland (four years before the first Blackberry). Do you remember how those mobile phones with Internet access looked at this time? The “Internet speed” of the Nokia 9000 Communicator was 56 Kbit/s.

Nokia 9000 Communicator

Nokia 9000 Communicator

It seems a century has passed since then, and this is the timeframe in which IPv6 was created. It is a truly old fashioned technology. Modern smartphones are already working with the fourth generation of the mobile Internet, whereas one level higher in the OSI model nothing has really changed. Obviously, the mobile Internet and IPv6 showed varying developments. After 20 years, four percent IPv6 traffic is indeed embarrassing.

What went wrong with iPv6? ^

So what went wrong with IPv6? Obviously, the prophecies that the IP address shortage would cause the Internet’s demise were hopelessly exaggerated. Nobody seems to be really interested in the other advantages of IPv6, because higher-level protocols can step in where IPv4 falls short.

Those long IP addresses just look ugly, and managing IPv6 is much more complicated than managing IPv4. This is an important factor, because end users still have to deal with IP settings. Even in corporate environments, IPv6’s complexity means higher costs, while offering only limited benefits.

I am no networking expert, but I think IPv6 was a step in the wrong direction. The slow adoption is essentially a verdict of the Internet community. In my view, low-level protocols should be as simple as possible, and high-level protocols should be responsible for complex networking tasks. Different applications, such as video and audio streaming, VOIP, web applications, etc., have different networking requirements that can be better managed at the application layer.

The mere shortage of IP addresses didn’t really justify the introduction of a completely new protocol. It would have been enough to extend the length of IPv4 addresses. This solution once worked well with phone numbers. For instance, we could have just added country and city codes to IPv4 addresses, and the address shortage would be no more. With such a simple extension of IPv4, by now, the transition would have already been accomplished.

The extremely slow adoption of IPv6 probably means that it will never really replace IPv4. It is more likely that both protocols will be replaced with something completely new that will come more or less overnight.

This doesn’t mean that I don’t support the adoption of IPv6. I am one of those progress addicts who like change just for the sake of change. IPv6 offers few benefits, but it is better than IPv4. IPv6 is the best IP protocol we currently have, so we might as well use it.

IPv6 adoption strategy? ^

One thing is for sure, though: at the moment, no organization needs an IPv6 adoption strategy as Coffeen recommends. I felt my head shaking when I read this:

Does your organization understand how this might impact the services you offer? For example, is your competitor’s website available over IPv6 today, while yours isn’t?”

So I have a competitive advantage if my web server supports IPv6? Really? That is a strong claim, considering no end user devices are currently running on IPv6. In mv view, such assertions are highly dubious, and FUD doesn’t really improve the reputation of the protocol or of the organizations that benefit from its adoption.

The truth is that almost no website runs on IPv6. For instance, the leading cloud provider Amazon does not even support IPv6 for its EC2 instances (only Elastic Load Balancers are IPv6-capable). If there really was a demand for IPv6 in the Internet industry, the most aggressive and forward thinking infrastructure provider would certainly support it. Of course, it doesn’t look much better at other providers. Microsoft also doesn’t see IPv6 support as a competitive advantage for Internet services, as Azure runs on IPv4 only.

So how is the situation in your organization? Do you have an “IPv6 adoption strategy”?

Want to write for 4sysops? We are looking for new authors.

Read 4sysops without ads and for free by becoming a member!

0
Share
25 Comments
  1. Nic Zarrilli 6 years ago

    IP addresses shortage is only relevant to devices connected directly to the "internet". Who does that nowadays? We all have a router which NATs our internal IP subnet to a single IP address on the internet. I think this is the major reason why the IP shortage as envisaged hasn't happened and therefore making IPv6 less likely to be implemented.
    The natural conclusion is that a new protocol is only required to router-to-router connections really. We are very happy with our IPv4 set up for the day to day stuff; simple to understand and to manage.

    0

  2. Nic, you are right, NAT is the main reason why we didn't run out of IPs yet. However, even with extensive usage of NAT, some countries might run in to problems soon (China for example). But I think if those countries move to IPv6, the rest of the world still doesn't need an "adoption strategy" because tunneling will ensure that the different IP networks stay connected.

    0

  3. Thomas Schäfer 6 years ago

    How about, adding AAAA to 4sysops.com ??

    Instead of spreading unjustified pessimism?

    Since 2011 the proportion of IPv6 has more than doubled every year!

    Waiting for a wonder is no solution "that will come more or less overnight."
    Upgrading the internet is hard work.

    0

  4. Thomas, give me one reason why I should support IPv6 on 4sysops?

    I think instead of spreading numbers that have no relevance you better tell the people the truth. What do you say about the 4% traffic share of IPv6 after 20 years? If we count the number of internet devices that run on IPv6, we would probably not even reach 0.01%. Any comments on these numbers?

    Upgrading the Internet is not hard work.The introductions of 3G and 4G were no big deal and happened more or less overnight. "Upgrading" the Internet with a technology that nobody really wants is hard. That's true.

    1+

  5. Thomas Schäfer 6 years ago

    You compare the small mobile networks with the internet?

    The mobile network still use also gsm (2G), The Introduction of 4g took in Germany 4 years and is not finished yet. (e.g. volte)

    Why you should add IPv6 to your own website? Then you have a better overview and your own statistics.
    Because you don't have IPv6, you have to trust google or apnic:

    http://labs.apnic.net/
    http://www.google.com/intl/en/ipv6/statistics.html#tab=per-country-ipv6-adoption

    In Germany a lot of customers are very unhappy with cgnat (ds-lite), their only chance to get back real internet is, that all switch to IPv6.
    Germany is only one example. The problems with cgnat are worldwide.

    0

  6. Thomas, no, I compare the huge mobile network with the incredible small IPv6 network. Don’t you think it is amazing that the introduction of 4G only took 4 years considering that after 20 years there are still no end users devices running on IPv6?

    I don’t need an AAAA record to determine the share of IPv6 traffic that would arrive at 4sysops. It would be next to zero. If this is the only argument you have for website owners, it doesn’t look good for IPv6. The problem is that IPv6 evangelists often give such dishonest advice. This only hurts IPv6 adoption because people start associating IPv6 with Internet snake oil.

    NAT caused problems in the beginning, but now it works quite fine. If there really were ISPs who are unsatisfied with CGN, they could have introduced IPv6 in 2006 when Windows Vista came out. Obviously, nobody (except a handful of IPv6 evangelists) really wants IPv6 on end user devices.

    0

  7. Thomas Schäfer 6 years ago

    You should not ignore the facts. There are "end" users using IPv6. (more than 10% in US and DE)
    It seems that you have no idea of the complexity of the whole internet. There are dependencies between hundreds of manufacturers - in backbone, access and end user equipment.
    You argue with Microsoft. Of course MS supports IPv6 since vista, but it still doesn't support it for xbox one(or just in a very ugly way). Microsoft is just one company with these contradictions inside.
    Of course not everybody did start 1998 with ipv6. From the first implementation to support the whole chain - that is a long way. And now, where we see countable success, you say please stop?
    We wasted to much time in nat/stun on IPv4.
    By the way - also mobile ISPs implement IPv6.
    Switch on IPv6 and you will see, how big "next to zero" is.

    0

  8. Where did you get this number of 10%?

    I think you misunderstood me. I don't say at all "stop." As I mentioned in the article, for me moving to IPv6 makes sense. However, my point is that the vast majority of the Internet community has a different opinion on this. Thus it doesn't make sense for me to ignore this verdict. I mostly wrote the article because for me as an independent blogger it is important to counteract the spreading of FUD. I think IT pros should make their decisions on real facts and not on the pipe dreams of evangelists.

    1+

  9. Paul 6 years ago

    Michael, do you actualy read the post Thomas write's?
    If so you probably have clicked on the links he placed 4 posts above, showing that United States and Germany have +10% and around 6 countries +5% IPv6 adoption.

    Besides that, your argument "I don’t need an AAAA record to determine the share of IPv6 traffic that would arrive at 4sysops. It would be next to zero." is just a bit of saying your own product is the best product. There is no way to prove your right.

    But i agree, the adoption is not going quick. But now the IPv4 adresses have no empty blocks anymore it will speed up. The biggest ISP's in the Netherlands seem to release it next year.

    0

  10. Paul, do you actually believe in the numbers from evangelists, especially if they also spread FUD?

    Google just measures the “availability of IPv6 connectivity” among their users. Does this mean those end user devices run on IPv6? I don’t think so. There are probably quite a few ISPs and organizations that have web proxies and NAT routers which connect through IPv6 to the Internet. I am travelling a lot, often work in coffee shops with different ISPs and have about 20 SIMs from different mobile providers. I must have about 100 WIFI hotspots in my connection list. Thus far I didn’t seen one public IPv6 address on my laptop."The only statistics you can trust are those you falsified yourself."

    But even if those number presented by Coffeen do have any meaning, it is quite obvious that he draws an unrealistic picture about IPv6 adoption by hiding the real interesting numbers. I am pretty sure that he is quite aware of how things really stand. This is my main criticism. Such people should not speak for IPv6.

    Let’s hope IPv6 adoption really speeds up now despite all this snake oil chatter we are already hearing for years.

    0

  11. Thomas Schäfer 6 years ago

    Is there a SIM from t-mobile US in your pocket?
    I have only 4 SIMs, two of them support IPv6.
    But it is useless for me to discuss - in your point of view I am an evangelist.

    You wrote:

    "Google just measures the “availability of IPv6 connectivity” among their users. Does this mean those end user devices run on IPv6? I don’t think so. "

    I know, they do. Of course to make statistics about the two protocols is not easy. ( NAT, Privacy Extensions, Proxys).

    Why do I believe that google-figures are correct?

    Because I see it in my (much much smaller) environment too.
    Logfile in apache, remote-worker via ssh. Normal people (no IPv6-evangilists) use IPv6 without knowing it. (The IPs I see come from kabel deutschland (a vodafone doughter), m-net (a regional ISP in Munich, and from Deutsche Telekom (the biggest ISP in Germany).
    I have had also seen IPs from other parts of the world - one student worked from Japan via kddi with an IPv6-source-address.

    There is a difference since 2011 and now, in 2011 IPv6 was only an noise in per mill, now you can measure it in per cent.

    0

  12. Thomas, you have to distinguish between evangelists and believers. Evangelists only tell you what serves their goals. Believers only see what they want to see.

    I have been using all German providers you mentioned in the last few months and they didn't assign IPv6 addresses to my laptop.I would have noticed this because I often work with VPN and with my current configuration this wouldn't work over IPv6.

    Because of the reason you mentioned, it is not just not easy but impossible to get data about IPv6 usage on end user devices my measuring traffic. Admins who give at least a bit about security don't allow end user devices to access the Internet with their IPs. Google is quite aware of this problem and this is the reason why they just talk about the availability of IPv6 connectivity.The only way to get reliable data about IPv6 usage is to ask providers and organizations. Measuring the traffic on central routers or web servers doesn't tell you much.

    I wish IPv6 replaced IPv4 10 years ago, but now we need something better. IPv6 is an outdated technology. Its main flaw is that it doesn't make IPSec mandatory. Snowden made it obvious that we now need mandatory encryption at the IP level, so all Internet traffic is encrypted by default.

    0

  13. Thomas Schäfer 6 years ago

    You must have been in a second Germany. You wouldn't have noticed IPv6, as long your vpn works with all kind of firefalled and natted IPv4. IPv6 is in Germany mostly implemented in combination dualstack or dualstack lite. A good vpn would also work behind NAT64.

    I accept your statement: "I am no networking expert..."

    Also your statement "Admins who give at least a bit about security don’t allow end user devices to access the Internet with their IPs." looks very strange to me. You have totally different understanding of the internet.
    You argue for encryption with Snowden. Just answer the simple question:

    Who has the keys?

    0

  14. Richard Ward 6 years ago

    2011? What about those of us who've known IPv6 for a decade and still barely use it? The last time I set up IPv6, it was to be used on another aging protocol: IRC.

    0

  15. Thomas, last time I checked there was only one Germany. Didn't we just celebrate the oneness? But then I am no expert on historical matters either. 😉 Anyway, I hope you are right. I would love to blog about all the problems admins will run into once they start deploying IPv6 (like malfunctioning VPNs).

    Richard, IPv6 is old, but not so old. If I remember it right, IRC was already dying when IPv6 was invented. Sometimes I wish developments in the internet wouldn't be so incredibly slow.

    0

  16. Tom 6 years ago

    Michael, I'm with you. I've been a network and systems administrator for over 20 years now. IPv6 is going nowhere. I suspect it will take a newer/better protocol with more benefits and easier implementation, before the work involved with moving away from IPv4 is worthwhile.

    1+

  17. Tom, you are right the benefits of IPv6 don't justify its implementation costs. I think a protocol that offers an easy way to encrypt internet traffic by default would be adopted very fast.

    1+

  18. Jean-Christophe Perrault 5 years ago

    I think IPv6 is a huge failure. l think the people who designed it completely failed.
    seriously go from memorizing about 5 or 6 numbers for each device on my network to now 64 Hexadecimal digits!!!! What were they thinking?
    Did they even have ONE admin to ask him what he thought about those crazy addresses?
    I know people want to blame the admins for not implementing but I have to learn new devices every week, I really don't have time to learn a new protocol that doesn't bring me any advantages. (PS: and I take care of my broadcasts with VLANS, thank you very much.)... and would make my work significantly harder. Not only can I not memorize IPv6s, I can't even write them on my standard cisco notebook, I have to turn it horizontal just to write one IP!!!

    I really hope someone more reasonable creates a new protocol or that just the ISP move to IPv6 and all the client stay on IPv4 with PAT and ISP converts the IPv4 with port to IPV6 addresses.

    I really don't see local admins starting to work with this horrendous invention.

    0

  19. Pablo 5 years ago

    There is not so much a shortage of IPV4 addresses as there is an unfair distribution and hording of them. Not sure how often you get out and about, but getting a Class C IPV4 subnet is downright difficult nowadays. Certain protocols like BGP often require whole class C subnets. For those who don't set up enterprise public facing networks, you can give whatever opinions you want. But some carriers seem flush with addresses and others say go ask ARIN.

    The problem is that ARIN should be charging for continued use of IPV4 addresses and should only be giving out IPV4 addresses to those who agree to simultaneously run IPV6 so that the transition is smoother. But ARIN has no authority, no teeth, no backbone, no IP addresses.

    0

  20. mem 3 years ago

    Yeah, IPv6 is dead. It will never replace IPv4

    All these people pointing at the mobile carrier networks using IPv6 really don't know what they are talking about. I've worked with all the US carriers at the service provider level and even some international mobile carriers, and their device networks are essentially closed loop bubbles...which is why they can "use" IPv6 to address the devices, but all those devices hit an IPv4 gateway on the way out to the "real internet"! It is just MIP IPv4 to v6 gateways. They could just as easily use IPX in their bubble with an IPv4 gateway. But the pie in the sky IPv6 proseltyizers mindlessly count up the number of cellphones in the mobile bubble network and declare "Look! Look! so many IPv6 devices"

    Puhhhleeeze. What a joke.

    0

  21. Thomas Schäfer 3 years ago

    All these people pointing at the...

    some examples:

    comcast (US)  is a cable provider

    Deutsche Glasfaser is no mobile provider

    Vodafone(Kabel Deutschland) has IPv6

    Deutsche Telekom (Germany) has IPv6 at DSL and mobile

    http://www.worldipv6launch.org/measurements/

    Thank you for the explanation of DNS64/NAT64.

    The IPv6-Internet  is no closed loop bubble, the IPv4-CGNAT-islands are closed loop bubbles and one way streets.

    But there are still fans of the old fragmented internet.

    https://www.potaroo.net/ispcol/2017-09/natdefence.html

    Despite the opinion of this old guy - IPv6 has reached 20% of the real internet.

     

    0

  22. Chira 3 years ago

    I completely agree.

    It seems that IPv6 has been designed by a combination of researchers/profs who, although technically skillful, have never touched a live production network.

    It's simply put an ugly and hard to troubleshoot protocol. No backwards compatibility with IPv4 and you had to rewrite a whole bunch of other protocols just to make the damn thing work. Just think OSPF. Such an elegant protocol, but for IPv6 you had to rewrite the entinre thing and to invent new LSA types and new CLI configuraton commands.

    For those who are old enough to remember or have read their industry's history, they will know that in 80/90s IP was just one of multiple protocols that could be used to build the nascent internet. It won out over its competitors (IPX, Appletalk, ATM come to mind) because it was robust, reliable and easy to imlement and troubleshoot.

    IPv6 is neither of those things.

    I remember the World IPv6 Adoption Day in 2011. Talk about much ado about nothing. That was already 7 years ago, and at the time I'm writing this comment, according to google statistics only 20% of internet traffic is IPv6 at the time I'm writing this comment (January 2018).

    The only people who have adopted IPv6 have only done so because they were forced by the IPv4 address exhaustion, and they had to be dragged into it kicking and screaming.

     

    2+

    • Author

      Thanks for the comment! You are right, the unnecessary complexity of the protocol is the main reason for the slow adoption. A protocol at this OSI level needs to be as simple as possible. All IP really has to do is to address computers behind routers. The rest can be done by higher level protocols. IPv6 was created by engineers who wanted to perpetuate themselves with a masterpiece. This didn't really work out.

      0

  23. Danilo 2 weeks ago

    IPv6 was supposed to be the future and yet 25 years later it has yet to be widely adopted. This is how IPv6 looks like in 2020: https://stats.labs.apnic.net/ipv6

    I think the main reason is the lack of backward compatibility. Apparently, when it was engineered no one thought about compatibility because back then IPv4 was not widely adopted yet.

    1+
    avatar
    • Author

      Interesting that those countries that have the most IPv4 addresses are leading. I think the problem of IPv6 is not backward compatibility. The protocol just doesn't have enough new features to justify the deployment costs. The shortage of IPv4 addresses can very well compensated with NAT. If IPv6 would have default encryption and authentication features things would look different.

      0

Leave a reply

Your email address will not be published. Required fields are marked *

*

© 4sysops 2006 - 2020

CONTACT US

Please ask IT administration questions in the forums. Any other messages are welcome.

Sending

Log in with your credentials

or    

Forgot your details?

Create Account