This has been discussed several times in the past. However, this is 2011, and things are slightly different now.

When does Linode plan to add IPv6 support?

I bring this up because according to estimations, we have around 10 days left before the "IPcalypse" and there are no more allocatable IPv4 addresses.

http://ipv6.he.net/
http://twitter.com/IPv4Countdown

(This is not an thread to debate the estimations or the implications of IPv4 depletion. Please stay relevant.)

I know there are various hacks to get IPv6 on Linodes (or any machine, really), but what we need is native support. Linode, being a primarily tech focused community, should definitely already be launching IPv6 support.

Also, Google is taking part in the Internet Society's "World IPv6 Day" and invites every webmaster to participate.

http://googleblog.blogspot.com/2011/01/ ... n-new.html

World IPv6 Day takes place on June 8, and I would like to participate without hacking my websites through a tunnel (which definitely doesn't have any benefit for speed, to put it lightly).

OMG - no more allocatable IPv4 addresses - EVERYONE PANIC ! ! ! ! !

There, I did my part (note the 5 double spaced explanation points for added emphasis on my seriousness and sincerity).

Wake me up when you can get coast to coast and continent to continent IPv6 NATIVE transport, then I'll start thinking about updating. Until them, meh.

@hart:

The more or less official word so far is that they're waiting until all of their data centers provide it. Currently, London and Atlanta are holding us up.

Once London and Atlanta get their butts in gear, how long it will take after that is anyone's (or caker's) guess.

@vonskippy:

You can't rain on my parade. I brought an umbrella!

_________________
Matt Nordhoff (aka Peng on IRC)

But without IP6, how can I put up a website that only 0.25% of the world can access ? ? ? ? ?

http://www.google.com/intl/en/ipv6/statistics/

As long as linode has IP4 addresses to hand out, why would you want to put a server on IP6?

Urgh, the point of this thread was to ask a specific question, not to host the weekly IPv6 argument... (Granted, the question is asked pretty frequently too, at least on IRC.)

_________________
Matt Nordhoff (aka Peng on IRC)

Wow, IPv6 is REALLY taking off isn't it. Who knew? Good thing I got into the IPv4 Sky is Falling So Lets Panic Now (IP4SIFSLPN) line early so I wouldn't miss my spot.

Eventually, when the regional IP registries start running out, ISPs will ONLY get IPv6 addresses. When that happens, ISPs and datacenters will probably be given very little or poor support (NAT across 2000+ customers) for IPv4.

For web developers, that has huge implications, as it will slow our applications and websites to a crawl, and many of those people wont have IPv4 internet access at all.

Also, Linode would probably dual stack the IPv4 and IPv6 (as most people will for the next few years). Our sites would live on both protocols accepting connections from each. That doesn't mean that upgrading to IPv6 would immediately make 99% of people unable to access our websites, it means the growing 0.25% (who will only have reliable IPv6 access) will be able to access our Linodes without having to be tunneled through a 6to4 service from their ISP.

Just a couple of thoughts...

1) I wouldn't put a lot of faith in the Google statistics linked. While I'm not arguing IPv6 adoption is low. the method Google uses has limitations.

2) A more interesting Google statistic, in my opinion, is the search for "Linode IPv6" returns about 20,500 results.

3) Telecity (Powergate) appears to have IPv6 transit - whether or not they offer it to their clients and/or requirements they may have is another question.

4) There's been global IPv6 transit for many years, not sure where that came from as that was the easy part.

5) Comcast, a large US broadband ISP, started conducting IPv6 trials in 2010. Why is this important? The other side of the equation, user accessibility.

6) You can make all the jokes you like, and some are well earned, but the fact still remains the IPv4 address space has limits and we can see those limits a lot easier now.

I personally would definitely like to see Linode move forward with an IPv6 adoption plan. Its been talked about for quite a while, why not start the process?

All I am saying is that exhaustion or not, IP4 will continue to work for years to come. IPv6 right now is still a playground that 99% of the internet can't access. You want to play with IPv6, good for you, I do too. But I just think that making assertions about how critical it is that linode jump on the boat is getting way ahead of reality.

It will, but I can't emphasize enough... When we're out of IPv4 addresses, every new internet connected device will ONLY have an IPv6 address. Unless your server can say hello on IPv6's terms, you will miss out on an increasing amount of traffic.

It's not that I want to play with IPv6. It's that I run 1 website in particular that gets much of its traffic from smaller countries. When their regionally allocated IPs run out, every new device from that point on will not be able to access that website, or any other IPv4-only website for that matter. As internet connectivity in smaller countries grows (exponentially), that's a lot of traffic I'm throwing right out the window.

You don't go to a party after it's over with several cases of beer. The longer Linode waits, the more and more like that scenario Linode will face. I, for one, will not wait for Linode. Any traffic lost due to the IPv4 to IPv6 transition is a loss in the money that pays for my Linode in the first place. If I don't get a solid IPv6 time table soon, I'm taking my business elsewhere. Rackspace Cloud, for example, is one such competitor with solid plans in place to enable IPv6 within early 2011. (While Linode simply blames the Datacenters.)

Unallocated IPv4 space and Unused IPv4 addresses are TWO completely different metrics.

When the Unallocated IPv4 space is finally assigned, the existing 4 billion IPv4 addresses don't stop working, don't disappear, don't explode/implode/whistle dixie.

The only (keyword ONLY) people getting the short end of the stick for the next several (keyword several) years will be the 3rd world nations that waited until now to join the rest of the 21st century on the internet.

Personally I couldn't care less if the vast number of new North Korea ISP's are stuck in lonely IPv6 land almost all too themselves. Of if Zumbagwagbagawiin's can't get enough (or any) IPv4 addresses.

The most pertinent question here (and it hasn't been asked) is what's Linode's pool of IPv4 space? In other words how many new VPS's can they spin up until they're out of IPv4 addresses. Then and only then will Linode customers have to worry (and then only if they have 100,000 addresses left and you're wanting the 100,001 VPS spun up).

In the meantime, the sky is still blue, water is still wet, and rushing to IPv6 (especially under the fallacy that if you don't it will cost you business) is still a huge Chicken Little adventure.

For my linode in Fremont, a tunnel via HE is sometimes quicker than native IPv4 (strange as that may sound) because cross country links using the HE IPv6 backbone seem to be slightly faster than their IPv4 backbone.
eg pinging from a machine in NYC to my linode:

linode ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4004ms
rtt min/avg/max/mdev = 83.175/83.328/83.792/0.235 ms

--- linode.ip6 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4006ms
rtt min/avg/max/mdev = 71.224/73.059/77.342/2.417 ms, pipe 2

Similarly from my linode pinging www.kame.net (in Japan)

ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4001ms
rtt min/avg/max/mdev = 123.636/132.833/142.356/7.228 ms, pipe 2

--- ping6 statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4005ms
rtt min/avg/max/mdev = 106.950/107.229/107.730/0.468 ms, pipe 2

If someone wants IPv6 on their linode that badlyy then a HE tunnel seems to be perfectly acceptable solution.

_________________
Rgds
Stephen
(Linux user since kernel version 0.11)

I never said that they would, and I don't think anyone else has.

My point was when IPv4 addresses run out, and all new connected devices have only 1 option (IPv6) and IPv4-only devices are in a world of hurt. Performance will severely degrade over tunnels that will most likely be offered by ISPs (many of which have their own technical limitations and security implications that most web developers have yet to consider), and as Google has proven time and time again in studies, speed matters to visitors.

I also agree, Linode's current IP pool needs to be considered. When it's out, it's out. No more IPv4 Linodes will be available unless someone removes a Linode from their account, which I imagine doesn't happen as often as people create Linodes (otherwise, they'd be going out of business).

Also, if you don't think that the US will have problems too, you are terribly mistaken. When your ISP takes away your dynamic IP and puts you on a NAT with a few thousand other clients, and your computer struggles to find 1 open port for internet access, you will feel the burn. If IPv6 isn't deployed fast enough, that IS what will happen. Do you think ISPs are going to turn down new clients just because they've loaned you an IP address and don't have any more? Nope. It may take the US longer to have the same problems, but rest assured that they're coming our way too.

Frankly, I'm not sure why this is such a big deal, or what the agenda is here in putting IPv6 down. So you don't think IPv6 is worth the effort? Fine, don't use it.

There are some of us, however, who realize that eventually IPv6 will be useful to have, and we'd like to be enabled on both networks so that we can play with it and make sure we have our configurations working well. And for that reason, it would be nice if Linode offered it as soon as they can reasonably provide it.

There's a big difference between you saying "eventually" and the OP saying OMG!!!! IPv4 runs out next week!!!!