In reply to a post by RobertoS:

As I replied to awoodland, if I understand NAT64 and 464xlat they look like the answer.

So are you in favour of CGNAT on IPv4?

Bob, 464xlat is a solution for existing CGNAT networks. It's used when migrating IPv4 CGNAT to IPv6 in a smoother fashion.

http://www.blubrry.com/ipspace/1729181/464xlat-expla...

NAT64 without 464xlat breaks some protocols. Try a SIP call via it and see how you do.

For residential CPEs dual-stack is the way to go. It's what people are already doing worldwide, from Andrews and Arnold to Comcast.

I'd not looked into CG NAT until it came up recently in relation to BT, although I'd vaguely heard of a Plusnet trial months ago. NAT64 and 464xlat were new to me in awoodland's first reply. So I am in no way qualified to answer your question.

CG NAT seems to have drawbacks. Dual-stack I've covered in this post.

I don't profess to be deeply knowledgeable about the various protocols. What is clear is that dual-stack without CG NAT ignores the fundamental issue - the exhaustion of IPv4 addresses, and with CG NAT is a horrible mess.

What is needed is something akin to DNS servers, (perhaps in the ISPs' routing systems?), or better still a modification to all DNS servers, that can do linkages between IPv4 and IPv6 addresses when the sender and receiver are not the same.

Surely that isn't beyond the wit of systems programmers? is it inherently more complex for an ISP to implement than CG NAT? I would have thought the two were roughly similar.

In reply to a post by RobertoS:

I don't profess to be deeply knowledgeable about the various protocols. What is clear is that dual-stack without CG NAT ignores the fundamental issue - the exhaustion of IPv4 addresses, and with CG NAT is a horrible mess.

I think what's critical is that every user has their own IP address. CG NAT on it's own does not offer that.

However, I'd fully support users sharing IPv4 addresses, but only if they have a dedicated IPv6 address as well, so that they can have ports listening to the internet if they so choose.

That makes sense Oliver, but still leaves the way open for providers to avoid the issue for ever.

Pure IPv6 needs to be standard and IPv4 consigned to the dustbin as soon as possible. There doesn't seem to be anything even remotely resembling any such intention.

I completely agree. And I find BT's CG NAT roll out not only a backwards step in the absence of a clear (public) IPv6 strategy, but also pretty contemptuous given that their customers seem to have gotten no warning, even though it has the capability to break things.

In reply to a post by RobertoS:

In reply to a post by blueacid:

One supposes that dual-stack could be later twinned with carrier grade NAT on ipv4 to achieve this... It's one solution at least and might explain BT/Plusnet paying that idea attention.

See my reply to yarwell. It's lunacy to have dual-stack at the user end.

Not really, it's the cleanest technical solution possible. What is lunacy is the ISPs and hosting providers who've got their heads so deep in the sand that they've yet to even try a technical trial, let alone a full scale roll out and have so far shown little or no signs of any willingness to do anything but a braindead sleepwalk into a horrible technical mess.

It's also lunacy that hardware vendors see IPv6 as a chance for another shakedown of their customers rather than just an (automatic) firmware update and it's lunacy that people like VM, BT, TT Sky etc. don't use their homogenous install base and considerable commercial leverage to push out a firmware update that effectively "flicks a switch" overnight.

Back on track: If the goal is not to run dual stack at the home user's property there are ways to achieve that, with caveats depending on the application. NAT64 can be run at an ISP level which works well enough as to be usable for general browsing. You can also run an HTTP proxy (which can be automatically configured, e.g. with WPAD) and browse IPv4 only sites from IPv6 only hosts.

In reply to a post by MrSaffron:

The key is for move devices like games consoles to support ipv6

If TF2 had hats for IPv6 connected players....

In reply to a post by RobertoS:

I don't profess to be deeply knowledgeable about the various protocols. What is clear is that dual-stack without CG NAT ignores the fundamental issue - the exhaustion of IPv4 addresses, and with CG NAT is a horrible mess.

It's a transitional state and the normal process for a transition is to run the old and new systems together for a time until the new one has been comprehensively adopted.

It's definitely less messy than either one or two layers of NAT.

EDIT: Incidentally, Bob, how do you propose NAT64 or 464XLAT don't ignore the issue of IPv4 exhaustion without putting customers behind asymmetrical NAT and causing the same issues as CGNAT?

Edited by deleted (Tue 07-May-13 20:05:20)

What about the paragraph that follows the one you quoted?

In reply to a post by RobertoS:

What about the paragraph that follows the one you quoted?

This?

In reply to a post by RobertoS:

What is needed is something akin to DNS servers, (perhaps in the ISPs' routing systems?), or better still a modification to all DNS servers, that can do linkages between IPv4 and IPv6 addresses when the sender and receiver are not the same.

That would simply give site operators another excuse to delay implementing IPv6.

I agree with Ignitionnet- the current system is the least worst way. Run the systems in parallel, when IPv4 addresses run out, site operators have a choice- implement IPv6 or see their hit rate inexorably decline.