So we have Google, who has probably applied manpower to the IPv6 transition of 100x the entire linode staff, says in six months they will be ready to test something for 24 hours, and people are DEMANDING that linode offer the same thing in production-ready form NOW!

Nobody in this thread has argued against IPv6 eventually. We are simply saying it is not a production-ready product now. The internet will limp by on IPv4 for the forseeable future, no matter how eager we geeks are to see it move on.

No one is "putting down" the idea of IPv6. I'm just saying "slooooooow down their chicken little" there is no need to panic, rush, or throw buckets of money on a problem that doesn't yet exist.

Linode offers a VERY nice service/cost ratio at the present, and I want them to know that most of us are not chicken littles that want to see ALL OF OUR cost raised because a few worry warts are ranting spend money on a ghost problem "now now now now now".


Since Linode seems to be VERY competently run by uber-geeks (and I mean that as a compliment) I can't imagine IPv6 is not on their top 5 must-have list, and that they're not making a very carefully orchestrated upgrade plan to add IPv6 to their stable of service JUST as soon as it's possible, affordable, technically doable, and NEEDED.

What Google and the other participants are testing on that day is to measure the impact on the client side of fully enabling IPv6 - not the IPv6 server side of the equation. Many sites that have implemented IPv6 on the backend have not published the IPv6 addresses for their primary sites due to concerns with how various client configurations are going to react. One of the goals of the project is to be able to more accurately measure that impact rather than just guess.

I would be delighted to join a beta for this at HE.

Ditto I'd go for a beta whenever it rolls around, I don't need it anytime soon but it'd be fun to play with. (I need to get out more).

_________________
Paid support
How to ask for help
1. Give details of your problem
2. Post any errors
3. Post relevant logs.
4. Don't hide details i.e. your domain, it just makes things harder
5. Be polite or you'll be eaten by a grue

This.

Linode, just like Google, will need to test it. Why not ask their large community of geeks via an opt-in beta available only to the datacenters with IPv6 connectivity.

We could help prepare the Library as well for users who will likely need to edit their resolv.conf (and I imagine even more, such as DNS) with new information. It would also allow us, willing participants, to find and document any issues with common server software, and eventually fixes as well.

How does waiting to do all of this help in any way? Oh right. It doesn't.

I'll ignore the silly ranting and just add that I would also be interested in native IPv6.

Do you have a link with some more info on this? I got my information based on checking the BGP table for the IP ranges Linode is present in. In this case, looking up london1.linode.com and putting that IP into http://bgp.he.net/

Googling for IPv6 Telecity Powergate will give you an idea of what providers have IPv6 transit available there, but as I put in my original post, whether or not they (Telecity) offers it to their clients is an entirely different matter.

my own 0.02. Im quite keen to see ipv6 native on the linode platform.

But, im also quite happy with 6to4 transition. I mean we all have static ipv4's anyways, so a big old /48 is perfectly acceptable to me.

However, if the IDC's are a problem when it comes to connectivity, why not go the tunnel broker route? I've seen a few chunks of free software out there that provide tunnel-broker like connectivity. Theres an integration hassle, which is probably going to be just as much effort as coding ipv6 into the linode portal once ipv6 becomes available at all their IDC's - so take that with a grain of salt.

But even without that, if linode were to host a 6to4 anycast router at their IDC's, then i cant see why people would hate being on 6to4 really?

Update: Sorry, what i meant with the tunnel broker bit was for linode to provide a tunnel broker service integrated into the linode portal.

That sounds far-fetched. Major ISPs already own a very large chunk of IPv4 addresses for their existing customers, and that address space isn't going anywhere. The only reason ISPs would NAT 1,000 customers on the same IP is if they somehow ended up with 1,000 times as many customers as they currently have. That ain't happening in the USA, since the majority of Americans are already online.

What's more likely to happen over the nest decade or so is that the ISPs will start NAT-ing customers who are on the cheapest plan, e.g. "DSL Lite". Those folks are the least likely to care. The more expensive plans, however, will probably remain NAT-free.

Do I like the idea of an NAT-riddled Internet? Of course not. But unrealistic claims of urgency don't help, either. Existing IPv4 addresses ain't going anywhere. Given how slowly some organizations move, it'll probably take 20 years for the world to transition out of IPv4.

Hybinet, your statements are easy to make when you think of the Internet as just a "series of tubes." Realistically, Verizon, the 4th largest ISP in the US last I checked, not only has high speed internet customers to allocate IP addresses to, but also their datacenters and their wireless customers. Yes, my phone has its own dynamic IP address. My two wireless cards also have their own dynamic IP addresses. Even when they're offline, they're still assigned to me. What I don't want is for that EVER to not be the case, especially considering the prices I pay for internet (especially from Verizon).

Me, just 1 person, takes up 4 IP addresses all the time. My home connection, my phone, and my 2 wireless cards.

It's not just about computers coming online. More and more smartphones are sold every day. More tablets with 3G/4G internet. More laptops with either wireless internet built in or with a dongle.

In addition, internet saturation in the US is around 78% last I recall. As computers, smartphones, laptops, and tablets get more affordable, the scarcity of IPv4 addresses, I believe, will be far more severe than you would like to believe.

And I'm really tired of people saying "Existing IPv4 addresses ain't going anywhere." I know that. I'm not saying they are. If you have to say that, you're not reading the entire conversation, you're just selectively responding with (in your head) clever (and redundant) remarks.

Few of those devices actually need a static IP address. ISPs can NAT or 6to4-tunnel the hell out of smartphones and wireless dongles, and only a small minority of tech-savvy customers will care. The situation is obviously suboptimal compared to one-IP-address-per-device, and I would hate to pay extra for my own IPv4 address(es), but many more devices can be accommodated without using additional IP addresses, and it will be done if necessary.

It's alarmism like the following that I disagree with:


This is pure speculation. Although deploying NATs and/or tunnels in a large scale will have hitherto unknown performance and security implications that need to be worked out, the issues can and will be worked out. I see no evidence to support the claim that the average customer's online experience will be "severely degraded" during the transition from IPv4 to IPv6. Maybe it will add a few more milliseconds of lag when you're playing your favorite FPS, but if you have a problem with that, you can always pay more to have your own IP address. There's nothing new to the story; it'll be just another chapter in the history of ISPs ripping us off at every opportunity.

Do you know much about port and connection limits per IP address? I highly suggest you research how sharing 1 external IP address to even a handful of people can affect website and software compatibility, performance, and security. Then multiply the not only potential problems, but highly likely problems on a very large scale (especially to, as you suggest, lower tier internet users).

Also, Smartphones need their own IP addresses for video calls, which are increasingly becoming popular. Adding a tunnel or proxy to very time-sensitive traffic can significantly degrade the quality, even if it's only a few ms of extra latency.

Yeah because that's "hardly" done now at all right?

The entire campus (of a west coast uni) has thousands of desktops, hundreds of WAPs, and hundreds of public workstations all behind a handful of public IPv4 addresses - so how (exactly) does NAT affect our compatibility, performance and security? I have ZERO problems with SKYPE (to Japan and Germany), UrT (a FPS game), VPN (to and from my hospital/lab/home office in Colorado), and banking/online bill pay. Perhaps YOU need to research how sharing 1 external IP address has practically ZERO impact on most computer users.

As to video cell calls - earth to George Jetson - where is that becoming more popular - Hollywood?