Originally posted by: Nothinman
If two ISPs, a and b, have only NATed access then users from ISP a can never directly communicate with users from USP b
Exactly. And any company with a decent security policy would have restrictions like that already. Home users are another issue, but you don't have to stick yourself behind a device doing NAT if you don't want to.
No, you totally missed the point - if IPv6 is never adopted, then IPv4 addresses are going to run thin, costs for those addresses are going to rise due to scarcity, or they will simply go into rationed allocation, and most likely, ISPs that are serving consumers, of which cost is a primary factor, will simply stop handing out publically-addressable IPs. That has already happened in places, and unfortunately, unless the IPv4 address crunch is removed via moving towards IPv6, it will continue to get worse.
So eventually, home users are forced behind their ISPs NAT devices, and have "no way out" to the real internet. Of course, their ISP can set up a transparent proxy on their internet network, and use their singular "real" IP address to proxy HTTP requests to other commercial web sites, and therefore, those ISPs will tell their customers to "shove it" when asked about the problem, because those customers can still access microsoft.com and google.com.
Think about it - that's a very real, and very scary, possibility. If that ever happens, then that means that the internet is dead, replaced by some psuedo-internet that is not much more than HTTP-based interactive cable television. Instead of viewing shows, you look at web pages. But the effect is similar, and the entirely democratizing notion of freely-available, end-to-end communications, is gone. We are all walled off from each other, all trapped behind our own ISP-imposed "iron curtains" or "berlin walls".
I sincerely hope, that if something like that ever starts to happen, that enough people move to installing their own decentralized wifi meshes all over their neighborhoods, and set up their own internet. It won't be easy, but it's still possible - unless, of course, the FCC decides to take back and close down free unlicensed public access to the ISM bands. Then we're all screwed.
I already consider it a partial-failure of the IPv4-based internet, that in most homes today, there are multiple people, each with at least one computer that they use as one of their primary communications devices to interact with the rest of the online world. It is already a breakdown in those end-to-end communications that the entire household has to deal with sharing a single IP address.
Couldn't you imagine the outcry, if every household, and every business, in the US was only allowed to have one single phone number per physical address? What if everyone in your household, had their own cell phone, but were still only allowed to share a single number? Every time someone called in, all of you, in different locations, would be prompted to pick up the phone, unnecessarily. Speaking of which, have you ever had to use a "party line" landline phone? I have. The entire section of the street is basically assigned a single line, with a single phone number. It's a crazy mess. That's the situation that IPv4 is basically almost in right now, and it can only get worse, the more IP-enabled devices and technologies that proliferate in the market today, and the only clear way to alleviate the problem, is widespread global adoption of IPv6 networking.
Edit: To try to bring this slightly more back on-topic, let me say this. The original question was whether or not AT would or should adopt IPv6 (I assume, in addition to, being on the IPv4 internet). That question veered off into discussions about the merits or lack thereof, of IPv4 vs. IPv6 right now, and the costs and benefits to businesses in general. However, a business, with many users, or many machines, would both have higher costs, and also I believe higher benefits, in the longer term, in moving to IPv6.
Anandtech, being primarily just a web site, with probably only a few machines in their hosted co-lo space, probably doesn't have a lot of need for multiple IPs, so therewould would benefit less from a switch from IPv4 to IPv6... BUT - it is for the same reasons, that the cost for AT to move to IPv6 would also most likely be lower, and therefore the transition easier, to move to IPv6-enabling AT. (Actual costs are I'm sure up to their upstream provider.)
But the *public perception* and positive PR gained by AT proving its technological leadership in the field, by adopting IPv6, would be relatively much greater, and would provide a PR boost to IPv6 deployment in general, I believe. Which is why, I believe that it should be seriously considered, and weighed against whatever costs they would incur from their upstream.