Slashdot Mirror
Vint Cert Warns IPv4 Users: 'Time To Get With the Program' (zdnet.com)
An anonymous reader quotes ZDNet:
Vint Cerf notes that the world ran out of IPv4 address space around 2011, some 13 years after internet engineers started sketching out IPv6, under the belief back then that IPv4 addresses would run out imminently. Since 'World IPv6 Launch' on June 6, 2012, significant progress has been made. Back then just one percent of users accessed Google services over IPv6. Now roughly a quarter of users access Google over IPv6. But Cerf noted that "it's certainly been a long time since the standards were put in place, and it's time to get with the program"...
The Internet Society's snapshot of IPv6 in 2018 notes that Google reports that 49 countries deliver more than five percent of traffic over IPv6. There are also 24 countries where IPv6 traffic is greater than 15 percent, including the US, Canada, Brazil, Finland, India, and Belgium. Additionally, 17 percent of the top million Alexa sites work with IPv6, while 28 percent of the top 1,000 Alexa sites do. Enterprise operations are IPv6's "elephant in the room", according to the Internet Society. Around 25 percent of all internet-connected networks advertise IPv6 connectivity, and the Internet Society suspects that most of the networks that don't are enterprise networks.
The Internet Society's snapshot of IPv6 in 2018 notes that Google reports that 49 countries deliver more than five percent of traffic over IPv6. There are also 24 countries where IPv6 traffic is greater than 15 percent, including the US, Canada, Brazil, Finland, India, and Belgium. Additionally, 17 percent of the top million Alexa sites work with IPv6, while 28 percent of the top 1,000 Alexa sites do. Enterprise operations are IPv6's "elephant in the room", according to the Internet Society. Around 25 percent of all internet-connected networks advertise IPv6 connectivity, and the Internet Society suspects that most of the networks that don't are enterprise networks.
We haven't "run out" of IPV4 addresses. Not even remotely so.
A good comparison would be land. There was a time, even within the last 50 years -- where one could (for example) 'stake out' land in Canada. You'd head to unclaimed land, put up your fences, work it and use it -- and in 5 (or 10? it's been a long time since I read up on this), the land would officially be yours.
This is closer to IPV4 realities, than not.
Why?
Because, IPV4 used to be *free*. You needed netblocks, you got netblocks. You request, and they were delivered.
Then they became non-free. Much like land in Canada, you can't just take it and use it, nope -- you have to buy it from someone.
A lot of that goes around, too. One corp selling to another. CorpA leasing to subscribers. ISPs selling additional IP addresses / month, for a fee.
If we had really "run out", I would have to WAIT to connect to the internet. Or, I'd be stuck behind a NAT device (I'm not), because my ISP had to aggregate clients because they had no free IPs.
Truth is, there's loads and loads of IPV4 laying around.
Otherwise, why would people be saying WE'RE GOING TO RUN OUT! for TWENTY FUCKING YEARS, and there's still a shit-tonne of IPs left.
Hmm?
Eh?
Hum?
Bah!
(And yes, SNI alone helped a lot... but that's not the point. Or maybe it is -- because, it's an example of "look -- there's gold all over the ground" and now "we have to dig for it, maybe we'd better use gold more wisely")
I bet in 2050, we'll still primarily be IPV4.
Nothing important uses only IPv6.
Spoken like a mere user. Those of us who've had to connect NATed enterprise networks via VPN, having to find common unused IP spaces, NATing around both ways to get machines from both ends to talk to each other, having to implement DNS zones, know just how wrong this is. IPv6 is a godsend, solving one hell of a lot of problems those of us actually working in networking have. Now, if only more of the management guys listened to us, we'd have moved on to IPv6 for quite a while.
Obviously we have to move to the larger address space, but IPv6 was invented by those most dangerous of engineers, those who think they're f'ing clever because they can make something complex and have lots of options.
When making the most core standard imaginable, that's like, the stupidest thing you could possibly do.
Many original core internet standards were widely adopted because they were simple for people to understand and program to.
204.92.16.108 etc is an example of this.
So in short, the IPv6 transition was made way more messy that it should have been, because of fundamentally incompetent design of the new standard.
Multiple ways of expressing addresses? Lots of special little address spaces reserved for this and that thing of the present day? Both of those are complete counterproductive bullshit. For example.
Where are we going and why are we in a handbasket?
$ dig tech.slashdot.org aaaa
tech.slashdot.org. 59 IN CNAME www.slashdot.org.
$ dig www.slashdot.org aaaa
(no answer)
This is half the reason why it's now the twenty-year anniversary of IPv6 failing to launch. IPv6 has now been around for longer than IPv4 (counted as the time between RFC 791 and RFC 188x) and it's still perpetually "the other protocol", the novelty thing that you use from time to time for a lark until you go back to the one that works. It's the Duke Nukem Whenever of network protocols.
The other half is that we've been told the IPv4 sky is falling so many times now that the response to any new claims are "oh god, is it that time of the year again?". For the vast majority of users, there's simply no incentive to switch, no matter how many times someone tries to scare them into it.