IPV6 is a funny one; it’s been destroyed by the ability to host multiple websites on the same IP, and IOT (another great BS term) stuff is mostly not publicly addressable, and when it goes online it accesses some private server.
When I joined Cisco in 1999, one of things that landed in my lap was the IPv6 development team. It was pretty small, so small I was encouraged to be economical with the truth when talking to customers for fear they’d think we weren’t taking it seriously.
The story back then was that if IPv6 wasn’t widely adopted within the next 3 years, the Internet would grind to a halt.
It was the same story when I left Cisco in 2006.
It’s the same story now.
It’s true that getting IPv4 address blocks is getting harder and harder (and more expensive). But it’s also true that huge swathes of the Internet, and corporate networks, are still IPv4 only, and will be for a long while yet. Meanwhile the IETF (standards body for the Internet) has invented more IPv4-IPv6 interworking techniques than you can shake a memory stick at, none of which have been adopted.
I can still get a virtual server for £5/month with a fresh IP, so IMHO IPV4 shortage is not anywhere near the horizon.
AFAICT, in more recent years at least, IPV6 has been driven by the hype of “IOT” (internet of things" which is currently as prolific as nuclear fusion. If you are building a product which has an RJ45 or WIFI on it, it is “IOT” 
There are lots of challenges in IOT. The most basic one is that one needs some sort of security (otherwise it might be trivial to trash everybody’s “internet enabled heating system”/etc) and this needs, hey, certificates, which, hey, expire! But a box which is supposed to run standalone can’t just go fetching new certificates because they could be spoofed (first you have to spoof DNS, which is probably easy, for a particular site). So the only way is to limit the internet access of this product to a private server; this gives you two things: a) you can charge €x/year for it and b) the product needs to store only 1 server certificate which will never need to expire because you control both ends. In the old days you just ran the heating controller behind an open port 80, with a username+pwd login presented, but the way “civil liberties” things are going, web browsers eventually won’t support HTTP, and the only other way is to develop a dedicated app which needs to be done for android, ios, windows, macos, and be maintained across versions of these, as a never-ending task, with no funding model because you sold the “IOT device” only once I am working on a product with some of this functionality (not domestic, fortunately) and one soon realises that so much of the stuff in the press is BS.
You left Cisco in good time, from what a friend who worked there told me
johnh wrote:
electric aircraft
It’s already flying.
Graphene has been just around the corner for many years. Technically it’s here, but practically not so much. But when someone managed to produce it at a reasonable cost and volume, nothing will be the same anymore.
Peter wrote:
I can still get a virtual server for £5/month with a fresh IP, so IMHO IPV4 shortage is not anywhere near the horizon.
Existing ISPs have IPs and can assign them to customers. They can have “a lot” available for their use. However, there is, generally, in Europe, very little to no available IPs for new entrants (a company that would want to start business as an ISP) and for growth (if an existing ISP needs extra IPs), In Luxembourg, the historical telco already makes a public IPv4 an option that you have to pay for on consumer-grade internet access. Not expensive, but just so that only people that want one get one. The people that don’t understand what that option is are behind Carrier-grade NAT, and counting their home router, that is double NAT.
Peter wrote:
IPV6 is a funny one; it’s been destroyed by the ability to host multiple websites on the same IP, and IOT (another great BS term) stuff is mostly not publicly addressable
Living on IPv4 up to now has taken rather profound engineering efforts (things like TURN, STUN, “NAT firewall hole punching”) to achieve behind NAT what would be rather trivial with globally directly addressable addresses for each device. To take but one example, the non-ubiquituous availability of IPv6 has delayed VoIP for several years, and made it reliably possible only with a “trusted” third-party server to “bridge” the media stream (at least by assistance in hole punching, up to actually relaying the media stream). This was the single biggest factor behind the early successes of Skype; they just had the best “NAT firewall hole punching” algorithm, and were willing to run servers to do media relay (and to exploit other users having nice public IP addresses as relays). Although I must admit that true P2P VoIP (with the “server” being at most a username-to-dynamic-IP directory) was also significantly blocked by many state’s expectations around “lawful intercept” (that is, “phone tapping”). They absolutely wanted to get “phone tap” from the “directory server” and not from the “Internet access ISP”, in a way that is “not detectable by the tapped person”, so changing the directory’s answer for the username-to-IP mapping to the Police’s (or the directory server operator’s) IP was “not discreet enough”.
johnh wrote:
widespread IPv6 deployment, (…) (a) which will come first, if any and (b) whether I’ll live to see any of them (doubtful).
I would understand having that opinion for ubiquitous IPv6 deployment, but IMHO we have widespread deployment already. About a third of traffic to/from Google (let’s take that as proxy for “eyeball traffic”) is already IPv6. That’s “widespread”, if unequal across regions and countries.
You may well be right about the VOIP interception business, but I played around extensively with VOIP on a phone, and while I was eventually successful, it was such a hassle that I didn’t bother with it. There were major blocks to it working. The first was the battery life hit of always having to be internet-connected; necessary to receive incoming calls, which the GSM system deals with much more elegantly by the phone always listening, at low power, to the GSM towers signalling it that there is an incoming call. GSM is basically a far better system for voice. The second was setting up the always-on connection (using STUN etc) which was even less reliable than bluetooth on Mondays 
The bottom line however is that the telcos don’t want people to use VOIP because they make money from GSM calls, whereas mobile data tends to be “free”. I am sure that calculation is shifting, with free national and limited-roaming calls. IPV6 would still not give every phone a dedicated IP, and would not deal with incoming calls without the battery hit. It may come one day… and VOIP is probably the only way for strong end-to-end encryption. I don’t know if calls via telegram etc are encrypted but they could be. VOIP normally isn’t; the data is sent openly with UDP.
VOIP works great for outgoing calls (albeit only on very good 3G+ connections – far higher grade than is needed for voice, and few people seem to know why) and I have it set up on most devices I travel with. It presents my mobile # on the CLI. I think the telcos deliberately run mobile data with a long latency, to cripple its use for VOIP, while advertising great data rates. In the UK, you could buy special VOIP deals which actually worked.
So yes VOIP is another great future technology which is nowhere near, for a pile of reasons
Just about battery hit.
I run Groundwire on an iPhone. This app is able to use « push » protocol. Hence incoming calls are available as long as you have data ON (I never switch them off) and battery is not drained.
My VOIP provider is OVH.
All in all, for 1€ per month, I have a professional phone number, and great flexibility with it (possibility for example to define DO NOT DISTURB periods, voice message sent to mail, etc)
lionel wrote:
very little to no available IPs for new entrants (a company that would want to start business as an ISP)
There’s no better way to explain slow IPv6 uptake. The ones who had to do it simply have a highly negative business incentive to do so.
Not sure what you mean by VOIP. Everything is VOIP these days, including 5G. Personally, I use Whatsapp and Google Voice for all my international calls, so I never pay anything beyond the basic monthly subscription for my phone.
I would understand having that opinion for ubiquitous IPv6 deployment, but IMHO we have widespread deployment already.
You’re quite right, but what I really mean is when IPv4 goes away. Right now you can use IPv6 if both systems and the intervening network support it, but you still need IPv4 for all the legacy sites that don’t support it, or if your network doesn’t. The real question is, when can you have an IPv6-only device that can access all the things you expect it to.
IPv6 development started in about 1990, before NAT had even been invented. Once NAT came along, it was no longer really necessary. Yes, NAT is a monumental pain, but in practice everything works with it now. In the early days there were a bunch of things, voice stuff in particular (H.323 iirc) that embedded explicit IP addresses in the payload. That breaks completely over NAT of course. So now everything uses a non-NAT third party – the remote access app AnyDesk is a good example, and so is WhatsApp.
