100 days of IPv4 left
infoq.com
infoq.com
The real key problem has always been that there's almost nothing to do in IPv6 space. I've got IPv6 connectivity (via 6to4 and HE tunnel broker) for more than two years. All I remember is watching that dancing turtle, visiting Google search over IPv6 (once or twice, maybe, because typing "ipv6.google.com" is inconvenient and pointless), read Python docs and updated various GNU/Linux distros from IPv6-capable mirrors. Oh, the only really useful thing is the ability to directly ssh/scp into any computer at my home without constructing additional tunnels to overcome limitation of having a single globally-routable IPv4 address per 6 devices.
It's not like most ISPs can provide IPv6 support easily, but even those who can see no point in doing so. NAT64+DNS64 is resourcewise worse than classic IPv4 NAT. Some (lots?) of clients will experience all sorts of strange client-side bugs with IPv6-only connectivity. And, more important, there're almost nothing to do in IPv6 Internet today.
I believe when hosting providers will start giving IPv6 blocks to every and each customer, ISPs will follow. Not the other way.
IPv4 has no problem with multiple network addresses assigned to one adapter, I do it all the time in both Windows and Linux.
Please, someone, make them stop telling that bullshit.
Only the first half of IPv6 address contains routing information. Okay, some routers may accept even /128 routes, but that's non-standard and one should not depend on that — the subnet is always /64.
So, the proper address space increase comparsion should be: 0000:0000:0000:0001:: vs 0.0.0.1, i.e. 2^64 vs 2^32.
Does anyone know of any hosting companies that offer IPv6 support? I have some clients on linode, but they don't support it either (http://www.linode.com/wiki/index.php/IPv6 - "Linode does not have native IPv6 yet.")
What you are suggesting is technically infeasible, and does not address the issue in any manner of significance. Congratulations, you liberated a few million IPs and delayed the IPv6 migration by a few days. Actually, since you freed individual IPs instead of entire blocks, you really didn't delay the migration at all but I thought I would be generous.
I don't have any clever or pithy quotes about this, so I'll have to wing it: If you have no expertise in the problem area, and you think you've come up with a simple solution that all the experts somehow missed, then you should probably assume you don't understand the problem and are wrong.
Of course, as has been pointed out already, even if all SSL hosting servers with one IP per certificate consolidated their certificates, it wouldn't help much because network routing relies on routing tables not being too big, so it is only useful if entire blocks can shared.
I'd say that deploying IPv6 widely will actually be easier than trying to salvage allocated IPv4 address blocks.