IPv6 Adoption Visualization
akamai.com
akamai.com
when you have 28 ipv4 address per 1000 people, it really leaves very little choice.
The central authority doesn't hurt but this is really an apples to elephants comparison.
That said, simply increasing the IP address field size in the header would have been a lot simpler and been easy to upgrade.
A big part of the IPv6 story is how they made so many changes there is a relatively steep learning curve to adopt. And NAT etc gave us another way not to.
It has an identical network model to v4: routing and address assignment work the same way. It works over the same L2 links as v4, and L4 protocols like TCP and UDP all work the same. DNS is the same. The socket API is the same. Link-local addresses exist in v4 too, and v4 even had router advertisements already before v6 was even out.
If they intended v6 to be as different as possible to v4, they wouldn't have designed it to work mostly like v4 with bigger addresses. They could've given it variable length addresses, or a locator/identifier split, or any number of other useful features that would've made it different to v4... but they didn't.
For IPv4 IANA eventually exhausted its supply of address blocks in 2011, when APNIC requested two more. That triggered their planned final distribution of one block to each of the five RIRs.
So, this chart is a distortion in the sense that nobody said "OK, Timor-Leste gets 4096 addresses" in some smoke filled room - at some point an LIR in Timor-Leste asked for a /20 maybe they're an ISP there or something, they got that from APNIC (the RIR for Asia Pacific) and APNIC in turn got their addresses from IANA as I explained above.
Each RIR gets to decide its own policies, most of them have a policy which says if you require IPv4 addresses today (a) you may just not get any, because they are exhausted and (b) if you can get addresses you might have to pay somebody else to release them and meet some qualifications to get those now-free addresses (but you can cut a deal so you're not paying for addresses somebody else ends up getting). They also tend to have a policy which demands you take IPv6 addresses, just to underscore that your correct way forward is not to hope that mathematicians discover more integers that were down the back of the sofa, but to just use IPv6 which has enough addresses for everybody.
IPv6 addresses are distributed exactly the same way, except they are of course very far from being exhausted because there are so many more of them and we've understood the "shape" of networks better this time.
You'd think that most everyone in both countries has a cellphone. In rural areas people may not have computers, but every farmer has a cellphone to order stuff from their cooperative or to check prices at market.
https://en.m.wikipedia.org/wiki/List_of_countries_by_number_...
This statement is still true China has far more internet connected devices than India. Although I will say it's a walled Garden, and India with new laws to control Internet related companies is moving towards walled garden.
Waiting for a response of the Union Government of India [0] [1] against a high court case in Kerala where the new laws are challenged.
If actions like this and promotion of pseudo science continue, India will soon be progressing backwards not forward.
[0] https://indianexpress.com/article/india/kerala-hc-notice-to-...
For the sake of being accurate, they're still rules and not laws. Unfortunately, the existing laws leave a lot of discretionary powers in the hand of the executive. But like you mentioned, the rules are being challenged in courts.
For many years in the early 2000s Ecuador lead the world in cell phone ownership because in most rural areas they had few landline phones due to the cost of infrastructure, so when cell phones came around everyone immediately jumped on board.
They also are very low cost, so they attracted lots of customers.
Those v6 numbers look promising though. Maybe one day we will be able to directly route between home networks again without dynamic dns and other ugly hacks.
If you are curious, you could always get a browser extension that shows you which IP version a given site was loaded on to see how many actually support IPv6, or just straight up disable IPv4 on your network for a bit to see what happens.
If I need to go through the rigamarole of tunnel brokering, BGP advertising, etc, I may as well just get my own block from ARIN. Vultr offers BGP brokering for a pittance, and they (presumably) have peering arrangements that handle the HE/Cogent debacle. I also want to try to get an IPv4 block, so I can enjoy SMTP hosting portability across ISPs. It would be nice to not be constantly harassed by Cogent sales staff. One of these days....
I was reading recently about whether you can get IP allocations and it seemed to me that ARIN basically requires that you show that you’ll be using some large percentage of addresses right away and keep growing that usage. Basically they seem to want to give them to ISPs or hosting providers. Can you can your own IPv6 block as a private small business?
There's otherways to qualify for an IPv6 delegation, but GP wants to be multihomed, at least a little.
Do you happen to know any guides on how to set up and use your own IP space, either with my ISP or with a cloud provider.
They'll most likely have rules/procedures for setting up BGP and what not. Look for 'bring your own ips' or similar language in cloud providers. Residential ISPs probably won't support it, but commercial ISPs all should, and may possibly help you with the ARIN procedures.
https://www.google.com/intl/en/ipv6/statistics.html#tab=per-...
We clock India at 66.06% and US at 59.37%
They should have had IPv6 with a header or config for IPv4 gateway(s) configured at the IPv6 boundary of your current network (could even be local / on device if needed). Want to talk to IPv6 - you go native. Want to talk to IPv4, connect using known prefix to IPv4 boundary gateway and it NATs you out.
The implementation / firewall changes needed for this are crazy, the config stuff also crazy, the number of layers of encryption I'm hitting (I'm IPv6 on a VPN using wireguard tunnels and browsing in SSL on my remote machine) seems nuts.
For me, the necessary firewall changes consisted of using "domain (ip ip6)" in ferm instead of "domain ip"; the general config is a few lines in radvd.conf plus a static IP on the LAN interface (and v4 requires that one too). None of that seems very crazy.
> the number of layers of encryption I'm hitting (I'm IPv6 on a VPN using wireguard tunnels and browsing in SSL on my remote machine) seems nuts.
This is two layers, one of which you introduced yourself and neither of which have anything to do with v6.
Those networks are Special because they run on custom hardware and can't take advantage of the work other people have done. But that's kinda their fault for not just doing their custom stuff on v6 in the first place; it's not like they didn't know it was going to be needed.
ARP is handled at the IP level rather than the ethernet level, and is done via solicited multicast.
Anycast exists, so OSPF is essentially part of the protocol. (Theoretically, you could dispatch at random, but you probably already built an OSPF tree for v4 routing)
The addresses are strictly heirarchial, with variable-length hostmasks -- this is even bigger than CIDR, as it's assumed subnets are actually masked off of a larger subnet
(There's also some boring stuff like the checksum being dropped)
These are all features of how IPv4 networks worked in 2004, but none of them were at the IP level.
http://www.faqs.org/rfcs/rfc1686.html IPng Requirements: A Cable Television Industry Viewpoint
Short answer: there was more stuff people wanted to fix.