Reply to post: 20% is not noticable

20% is not noticable

20% of anything generally is a rounding error for me. 20% faster, 20% cheaper, 20% more efficient.

I just checked my AT&T phone again, it is not roaming, and is assigned an IPv4-only IP 10.xx address. www.whatismyipaddress.com says I am coming through a proxy server(or perhaps carrier grade NAT maybe they don't differentiate). Never had a problem accessing things even with tethering.

I just spent 3 months in thailand, didn't happen to notice what my phone with a local SIM card was assigned if it was v4 or v6, but more than anything the latency(purely distance I imagine) killed performance, I managed to get somewhat better web performance tethering and going through a VPN and proxying my HTTP traffic through my co-located server in California, though not for geo-diverse sites).

Myself I have no interest in IPv6, I do not have any plans to deploy it on any of my networks, if it happens to get deployed in a transparent fashion upstream on my mobile or broadband connection I don't care since it is transparent (I do not make inbound connections to my phone, and any inbound connections to my broadband are done through a VPN established outbound to my co-located server so I don't need static IP and don't have to care about NAT that the ISP might be doing).

I figure if at some point I really need to support inbound IPv6 to the websites that I manage then I'll have that translation handled by the CDN (just like they already do things like SSL termination and TCP optimizations already).

The die hard IPv6 fans remind me of the modern web developers who want you to jump on the new thing just because.

I would put IPv6 up there along with Software defined networking(SDN) as something that can benefit large scale companies well(mainly service providers), but for 95% of the downstream organizations it provides no benefit.

IPv4 addresses are available, getting large blocks of them may be difficult, though every company I have worked for in the past decade hasn't needed more than a few dozen external IPs at most (even today not hard to get). Smart use of NAT and name-based virtual hosting(with good proxy servers - in my case Citrix Netscaler or F5 BigIP) goes a long way.

Now that SNI is pretty widely adopted that removes a large reason to need a big number of IP addresses on servers with multiple SSL certs(covering multiple domains so a wildcard cert wouldn't be sufficient).

Die hard IPv6 proponents say IPv4 is full of hacks and brokeness so we should upgrade because of that. The hacks seem to work fine, I suppose if you are directly suffering from those limitations then you should probably upgrade, myself, I really haven't had any pains associated with IPv4. For example the last time I was at a company that needed a site to site VPN with another company and we happened to have overlapping IPs so had to do 1:1 NAT -- that was literally 11 years ago).

Myself I don't do peer to peer anything and the VoIP solutions I have used seem to work fine with IPv4 and NAT etc.

So no pressing need to upgrade, I don't see a pressing need for many years to come(for folks like myself anyway).