5% of Google's traffic is now IPv6
google.com
google.com
° https://github.com/docker/docker/issues/2174
° https://github.com/docker/docker/pull/8947
° https://github.com/docker/docker/issues/3609
etc...
https://github.com/docker/docker/search?q=ipv6&type=Issues&u...
2.5% in December 2013
1% in December 2012
0.4% in December 2011
0.2% in December 2010
If that carries on then it's 10% in 2015, 20% in 2016, and 40% in 2017, with everyone switched over somewhere in 2018/19.
160% in 2019
Apologies for being flippant. Just making the point that you're extrapolating without spending time on the likely shape of that adoption.
It seems more likely that this is an S-curve, and that we'll continue to see IPv4 traffic for the next decade, maybe longer.
I'm sure that IPv4 support will continue for a fair while though.
It feels like it'd need to be a dramatic cost saving in order to kick in at 50% adoption.
For example, I find it easier to have an https-only web infrastructure. Too bad for old deprecated browsers who cannot support proper encryption (and therefore become a security risk for me, and a waste of time to support). Dual protocols (v4 and v6) take more time and money to manage. Initial training can be harsh. Some first-gen hardware was terrible. However, now that it's done, I can't wait to switch off IPv4.
ARIN is going to run entirely out of IPv4 addresses for general distribution in April of 2015.
It's not going to be long before you'll need IPv6 if you want to do anything peer-to-peer.
Only meant partly in jest. When they start to get really rare, the price of IPv4 blocks will keep going up, until some institutional holders of large blocks start selling. It could take quite a long time before it's really gone for good.
A couple of years ago my ISP dropped their usenet servers - not enough people were using them to justify the cost/maintenance effort. I expect we'll eventually see the same thing with IPv4; not for ten or twenty years perhaps, but once "everything" is using IPv6, it becomes a cost center.
Already seeing stuff like that happening in certain asian countries, where companies are only available over IPv6.
You give people a lot of credit. I work at a large ISP/Telco, we just got a new Director of "Core/IP"... in the first meeting the new director attended we started talking about IPv6 initiatives and the new Director interrupted us all and said "Wait. What's IPv6?"
(They were serious)
They hire managers and Directors who will do what the people above them say. The people above them have even less technical know-how, so it's not like they can vet potential candidates.
1. The color selections on the "Per-country IPv6 Adoption" tab are abysmal for those of us with poor color perception.
2. Since we're talking about replacing outdated technology, why use a Flash-based graph on the "IPv6 Adoption" tab?
"Google works, so it's not my connection at fault, it must be your site that's broken."
The big reason to support IPv6 now is IMO that (all else being equal) it's better to add v6 support when you roll something out than to introduce it later.
Our network guys seems to love the extra complexity, though.
It sounds easier than it is, the main issue is scaling. Need to connect to one other person for B2B? Trivial. Two? A little harder. Ten, twenty, fifty, ouch...
On the other hand, corporations tend not to have the same level of skill, and focus in terms of networking as an ISP, like a Comcast, or an AT&T. Additionally, they're not as incentivized to move everyone to IPv6, because often times, it's easier, cheaper, and more realistic to buy some IPs.
Moving to IPv6 at the office is a difficult project for most, as there is still a ton of enterprise gear that's IPv6-less. Even at that, enabling enterprise applications with IPv6 tends to be a massive nightmare. Lastly, the security implications of IPv6 are not well understood by most people in the corporate environment.
Was there maybe a new phone or console (or an OS update released to either) released?
There's a good chance a lot of this traffic consists of those bots testing Google's results ....
I actually do believe that this is real traffic. Here in Germany the number of connections with native IPv6 is rising constantly. At home all my devices are always using IPv6 when connecting to Google services because I also have native IPv6. Those 12.26% for Germany look pretty believable.
I'd tried some IPv6 settings in my router last year, and 6to4 worked pretty well at that point.
It is noticeably faster to use IPv6 with Google, something like 4ms vs. 30ms connection times. I'm not sure if it's a more efficient route, or underutilized routers handling the traffic.
United Kingdom: 0ms latency, 0.02% impact
Romania: -10ms latency, 0.01% impact
Bulgaria: 10ms latency, 0.03% impact
Japan: 20ms latency, 0.03% impact
All that I can divine so far is that 20ms appears to be the threshold at which your country will receive a red tint regardless of your adoption percentage.
Is there something I need to change on my end? I've got pretty vanilla settings. Thanks!
1. What kind of IPv6 connectivity do you have? If it's not a native deployment, it could be 6rd or 6to4. I believe that some applications/OSes will prefer IPv4 over non-native IPv6 when they have the choice, since the latter will be running through a proxy (and therefore will probably be slower).
2. It could be an OS setting. I think that most recent OSes will prefer IPv6 by default, but I could be mistaken.
3. Application settings could also be in play.
I installed a Chrome extension named "IPvFoo" that adds an indicator in the address bar to tell me whether I'm connecting over IPv4 or IPv6. When I'm at home (Comcast residential, native dual-stack), it shows me connecting over IPv6 a lot of the time, especially when I'm connected to larger sites; in fact, the majority of sites connect over IPv6 at least partially, since many popular services (Google Analytics, Google Fonts, some common CDNs for Javascript libraries) are available over v6.
Perhaps this will help you figure out what's going on: http://test-ipv6.comcast.net/
1. I do get an IPv6 address from Time-Warner... and it appears to be legit. (It's not the link-local address; it starts with 2605:e000, which is in the Time-Warner block)
2. But... it appears to be non-routable from the general internet. I think this is Time-Warner's fault. But maybe it's my router. I am not sure. Traceroute from http://4or6.com/traceroute?l=en gives me this. (I attempted to obfuscate my real address with XXXX)
#traceroute 2605:e000:xxxx
traceroute to 2605:e000:xxxx (2605:e000:xxxx), 30 hops max, 80 byte packets
1 2607:f2f8:1600::1 (2607:f2f8:1600::1) 1.653 ms 1.597 ms 1.583 ms
2 2001:504:13::1a (2001:504:13::1a) 11.753 ms 11.746 ms 11.732 ms
3 twcable-backbone-as7843.10gigabitethernet17.switch2.lax2.he.net (2001:470:0:2bf::2) 2.618 ms 2.610 ms 2.599 ms
4 2001:1998:0:4::11d (2001:1998:0:4::11d) 3.956 ms 2001:1998:0:4::11b (2001:1998:0:4::11b) 3.477 ms 2001:1998:0:4::11d (2001:1998:0:4::11d) 5.512 ms
5 2001:1998:0:8::83 (2001:1998:0:8::83) 4.090 ms 5.784 ms 2001:1998:0:8::87 (2001:1998:0:8::87) 6.542 ms
6 2605:e000:0:4::41 (2605:e000:0:4::41) 6.753 ms 5.032 ms 5.011 ms
7 2605:e000:0:4::5:7f (2605:e000:0:4::5:7f) 6.859 ms 7.042 ms 7.128 ms
8 * * *
3. Since a recent firmware upgrade, my router detects this problem, and sometimes (but not always) decides to set up a 6to4 tunnel. I am not sure how this works, exactly.4a. If the router does set up the 6to4 tunnel, I do have IPv6, but as organsnyder says, since 6to4 tunnels suck, the OS prefers to use ipv4.
4b. If the router does not set up the 6to4 tunnel, I have a completely non-functional IPv6.
Oh well, maybe next year.
The idea would be to set up personal VPN that would exit thru IPv6 and/or get anonymomized on a public IPV4 NAT. Would that make sense?
[1] Their London PoP is not accepting new users right now https://tunnelbroker.net/new_tunnel.php
[2] Two PoPs in London https://www.sixxs.net/pops/
If anonymity is your goal, then a VPS probably wouldn't be much help—you'd be traced right back to your VPS provider, so you'd still be hoping that they wouldn't divulge your information. Since a VPS is normally on a static IP, it'd be slightly worse than a typical dynamic IP from a residential ISP.