Comment by 10000truths
5 days ago
It's hard to adopt something that schools don't teach. I know someone who graduated from UCI with a CompSci degree with a specialization in networking, just before the COVID19 pandemic began. He recalled that the networking courses he took did not cover IPv6 at all, except to describe the address format (i.e. 128 bits, written as hexadecimal, colon-separated). Everything he learned about IPv6, he had to learn on his own or on the job. A standard that has been published for over two decades, heavily used for over a decade, and critical in the worldwide growth of the Internet, was treated as an afterthought by one of the premier universities in the US.
Obvious disclaimer: This is a sample size of 1, and an anecdote is not data, yada yada. I'm not involved in academia, and have no insight into the adoption of IPv6 in CompSci networking curricula on a broader level.
Meanwhile, I was taught and practiced IPv6 in 2003-5 in engineering school (France).
As of 2024, IPv6 deployment in France was >97% mobile and >98% residential due to not being required for obtaining a 5G radio license (and then v6 simply carried downward to being available on 4G) + every ISP that provides FTTH also providing v6.
https://www.arcep.fr/fileadmin/reprise/observatoire/ipv6/Arc...
Over here IPv6 JustWorks to the point of absolute boredom.
I was taught IPv6 in the mid 2000s too, in Italy.
But penetration there is just about 15% or so :/
Is it commonly used within small/medium/large businesses?
German situation is mostly/rarely/never. Small businesses have their DSL line where their cheapo router will announce an IPv6 prefix which almost all ISPs over here provide. Medium to large businesses usually have some braindead security policies that include switching off all IPv6 functionality in devices.
9 replies →
Tbh it’s is a huge PITA with little practical benefit. IPv6 is the Perl 6 of networking.
Many of the big benefits are things that don’t deliver anything that folks are lacking. You also need to understand how you fit in the overall universe more.
An example for a small environment: I've got the whole homelab on unique ipv6 range. Whatever VPN connection happens to another network, I'll never have range collisions or need any fancy rewriting. Also the DNS will point at a specific address on my network, never at a random 192.168.x.x in a network I happen to be connected to.
You’re not wrong, but I have been running complicated multi-site VPNs with a small homelab multi-subnet / VLAN setup for 25 years and still have yet to have a collision.
My home network is dual-stack these days, but because my IPv6 prefix is dynamically delegated by my ISP, I actually use site-private IPv6 addresses for all my internal servers and infrastructure.
The thing is though, I don’t even need IPv6. Comcast Business broke my delegation for six+ months and I literally didn’t even notice.
IPv6 tried to do way too much. The second system syndrome was strong. It’s no wonder folks are annoyed at the complexity, and as long as IPv4 continues to works for them, they aren’t particularly pressed to adopt it.
4 replies →
> never at a random 192.168.x.x in a network I happen to be connected to.
That’s a pretty good benefit, I hadn’t considered that!
Eh, I've been thus far unimpressed.
Part of it being that a lot of ISP's don't have static prefixes, they do get rotated pretty often and have no guarantee of CIDR size that you're going to get. By default my ISP will only give a single /64. You have to go out of your way to request more subnets and there's no guarantee that the ISP will honor that request.
It's really problematic to try and base a non trivial network setup, when you have no guarantee of how many subnets you can run. Today I've got 256. Tomorrow it might be 16. Or 2. Maybe just 1 again. ISP's can be weird when they smell monetization dollars in the water.
So I have to run a ULA in parallel to the publicly accessible networks specifically for internal routing, and then use a DNS server to try and correct it. Which works great! ...except when you run into this little niche operating system called Android. Which by default doesn't obey a network provided DNS server if you've got privacy DNS enabled. So if I've got guests over and I want them on a network in my place to access some sort of internal resource, then I've got to walk them through disabling privacy DNS.
Either that or I need to go out and buy a domain... for my internal network...and then get a TLS certification for my private internal domain.
I get how IPv6 can be great. But a lot of the advantages are also overhead I don't want to deal with.
Short hand is a good example; I've lost count at the number of times I've typo'd short hand addresses because my eyes skip over a colon. At this point I've gotten into the habit of just writing out the whole address, leading 0's included because the time saved from not making a mistake reading the address often faster overall then making mistakes with shorthand.
2 replies →
What about the benefit of there being enough addresses?
That particular benefit has no value if you still need to support v4.
It's almost a self-inflicted tragedy of the commons or reverse network-effect.
Adopting IPv6 doesn't alleviate the pain of IPv4 exhaustion if you still need to support dual-stack.
16 replies →
The widespread deployment of NAT and VPNs has counter acted the market forces that were assumed to make IPv6 appealing.
28 replies →
That is a collective problem, though, not an individual one. I have always been able to get enough v4 addresses for all my needs.
8 replies →
There’s zero benefit to you because the carrier is NATing you for other purposes.
They get better network management.
Enough addresses for what? Nobody needs or even wants all of their devices to have globally routable addresses.
3 replies →
Everyone who says this is obviously a web developer.
That's a pretty bold claim. IMO IPv6 is not hard at all, and delivers significant benefit when dealing with anything outside your local network.
I absolutely love the things that IPv6 delivers and employ it on purpose.
The world very clearly doesn’t revolve around what HN users “love”.
4 replies →
This is so right.
No One believes us on hacker News. It feels very gaslighty. I have never talked to an IT engineer in person that thought IP version 6 in the data center or in the corporate network was a good idea.
I recently passed the CCNA again and they really spend a lot more time on IPv6 compared to 15 years ago. It inspired me to go all in this time and configured my home network with a PD allocation from my ISP. I also came up with some fun labs and even got a IPv6 sage T-shirt from Hurricane Electric.
Did you have to do anything special to get the t shirt? I got the sage cert ages ago and they never sent my shirt...
No, but it did arrive several months later. Maybe they wait and send it in batches?
Any recommended courses? I'm a SWE and never felt compelled for the CCNA but my intersection with networking-related problems seems to continuously increase and I would like to up my game before getting in over my head at work.
I just bought the official exam guide found Neil Anderson’s videos helpful. One thing that bugged me a bit was they spent a little too much time on their WiFi, including the obsolete Airie OS.
This doesn’t hold up. Schools can’t teach everything, especially in a field where innovation happens in the workplace, not the classroom. Should I have learned about LLMs when I was an undergraduate 20 years ago?
This is just further proof that university educations are still not job training. The sooner we disabuse ourselves of that perception the better off society will be.
Higher education is about creating a breadth of knowledge, not specific marketable skills. CompSci is a research field, not job training.
If your friend wanted to learn specific job skills a technical college would be the appropriate setting.
I realize this misperception is perpetuated by the job market but I’m still not surprised at the education provided by UCI and don’t fault them for providing it.
They taught us, they also taught ipv4 in the old "separate address per host" way instead of jumping to NAT, but I think ipv6 is inherently more complicated than ipv4 for the average use case. It's not just a thinking shift.
Separate from that, deliberate decisions were made to make it a "clean slate" without consideration for existing ipv4 hosts. Guess they were hoping the separate stacks would go away eventually, but in hindsight, no way.
> ... but I think ipv6 is inherently more complicated than ipv4 for the average use case. It's not just a thinking shift.
IPv6 isn't all that complicated for most common use cases. Its fundamental concepts and rules are simple. It also obviates the necessity of the complicated workaround called NAT, without which IPv4 is impractical these days.
It's more like the imperial vs metric system debate. If the world hadn't seen IPv4, I believe that we'd all be using IPv6 without any complaints. The real problem is that IPv6 isn't taught well.
> Separate from that, deliberate decisions were made to make it a "clean slate" without consideration for existing ipv4 hosts. Guess they were hoping the separate stacks would go away eventually, but in hindsight, no way.
I'm not sure what to make of this. The presence of the IPv4 stack isn't what blocks the adoption of IPv6 - at least not technically. They can coexist on the same host and function concurrently without interfering with each other. It was designed to operate like that. The actual blocker is the attitude that people hold towards IPv6 - "We have IPv4 that works already. Why should we care about an alternative?". You can see that expressed on this discussion thread itself.
There is one crucial detail that the IPv6 detractors neglect - the scarcity of IPv4 addresses means that IPv4 address blocks are now heavily coveted and therefore subject to moneyed interests. That isn't very good for the health of the open internet, digital rights and equity. They're thinking about individual trees and losing sight of the whole damn forest. IPv6 isn't a solution looking for a problem. It's the solution for a problem that people simply ignore.
The IPv6 spec was being modified up through 2017. It has more kinds of addresses that behave in fancier ways, with one host having multiple. The very first thing you see with ipv6 is your nice memorable ipv4 addr replaced with a long hex string with some ::s thrown in. Local DNS is commonly recommended with ipv6 for that reason, which maybe is just some misguided advice because it sounds crazy. I guess you could assign and memorize ULAs?
NAT is technically complicated if you're looking inside it, but most people aren't, and for them it's really easier to think about. You've got a public and a private, and there's a very strong default that private isn't exposed. People screw up firewall rules all the time or routers have bad defaults, but it takes more deliberate action to publicly expose a port over NAT. Plus you don't need privacy addresses that way (introduced to ipv6 in 2007). I know "NAT isn't security" but for most people, it is.
Still not even sure what the accepted default firewall behavior is in ipv6, cause some people say "ipv6 lets any device do p2p by its own choice" and then when you ask about security, "your router firewall should always default-deny anyway," so which one is it?
> The presence of the IPv4 stack isn't what blocks the adoption of IPv6
It is. Like they say, most technical problems are really people problems, especially this one.
5 replies →
ipv6 would have been a breaking change anyway, just take the opportunity to push through any changes that they want to make
Helsinki CS masters had ipv6 20 years ago, but nobody listened at the lectures because all of our home LANs ran ipv4
You have it backwards, education always lags industry adoption. (*Assuming it's a software engineering-focused curriculum.)
Programs will teach Docker only years after it is adopted.
Same with AWS, JavaScript, etc.
If it’s not adopted by industry, it won’t be taught about in schools.
I got taught IPv6 in 1995. At that time they said it was super important because it would replace IPv4 within a year lolololol
I can’t think of any technology where mass adoption was driven by knowledge forcibly inserted into students’ brains by schools… if anything, adoption comes when people realize their out-of-touch curriculum is no longer relevant.
To be clear, degree programs have value, but it’s not in future-proofing students against needing to learn things after they leave school. Ideally it should prepare them and encourage them to do so.
>> I know someone who graduated from UCI with a CompSci degree with a specialization in networking, just before the COVID19 pandemic began. He recalled that the networking courses he took did not cover IPv6 at all...
I am not doubting you, but I feel this story is too hard to believe without adding further nuances...
MIT 6.829 teaches IPv6 since 2002: https://ocw.mit.edu/courses/6-829-computer-networks-fall-200...
In Portugal and other countries, there are subjects on Computer Science before College or University, and they teach it on High School...
The issue is that it’s not taught with IPv6 first. Networking courses do all kinds of stuff using IPv4 to demonstrate various protocols on top (e.g. http, tcp, icmp, etc).
Then there is usually a chapter on IPv6 that just briefly covers the differences.
I.e. the exercises all tend to use IPv4 as the foundation so people don’t practice v6
But TCP or HTTP don’t care about the underlying transport. They’re higher level protocols that are payloads to either IPv4 or IPv6. It’s irrelevant what the transport is when dissecting HTTP and very little time should be spent on it.
IPv4 is, for all intents and purposes, still the default transport. It’s also simpler than IPv6 in some regards. When teaching layer 3, it makes sense to teach both, and teach IPv4 first. Though I fully agree that they should be taught with equal emphasis. I don’t doubt there’s a good number of programs out there that don’t into sufficient detail on IPv6.
1 reply →
Well it makes sense, no one uses ipv6 anyhow. Most I know are waiting for ipv8.
I've been of the opinion this is one of those "the art advances one funeral at a time." A lot of people are married to IPv4 and its arcane warts and really, really do not want to deal with IPv6 even though most of the core concepts are almost exactly the same thing, except better. I can't imagine anyone who dealt with V4 multicast ever wanting to go back, and I bet they've memory-holed parts of V4 that simply can't be used anymore and so have been turned off for decades(RIP to RIP). Has anyone seen the automated address assignment in V4 ever work? The usual hint it even exists is that if you see one of those addresses it means something is messed up in your Windows host or the DHCP server died.
People complain about dual stacks and all that but with a modicum of planning it is minimal extra effort. Anything made in the last decade has V4/V6 support and unless you're messing with low level network code, it's often difficult to even know which way you're being routed. Network devices pretty much all support using groups of names or addresses and not hard coded dotted-quad config statements now, and have for a while. And that was good practice on V4 networks too.
Part of it is probably that remembering various V4 magic is easy enough to do but feels complicated enough to be an accomplishment. In V6, there is no point in doing most of that because the protocol has so much more automation of addressing schemes. But if you like those addressing schemes, V6 can do them even better. You can do all sorts of crazy address translation on either the network or host id portion, like giving an internal network a ULA that is magically translated to a public network prefix without any stateful tracking unless that is desirable.
I feel there is some analog to DNS in that regard, people who have gotten used to DNS don't give a damn about host IP addresses but some people seem to really like the idea of a fixed address statement. People also seem to be stuck on the idea that NAT creates some kind of security when that's really the stateful tracking that is required for many-to-few translations (thus making firewalls a common place to implement it), not the translation itself. Similar to certificates/pki versus shared keys, yes, one is more upfront effort but that's because it's solving the problem of the Sisyphean task that is the other.
edit: This all reminded me that we lived with dual stacks before, in the IP and IPX days, or DECnet, and that GE Ether-whatever, and those had less in common. IPX mostly died with Netware but it had a number of advantages that wouldn't be bolted on top of IP for years, some of which are present in IPv6. I rather liked IPX and had history gone differently that it used 48-bit addressing would be causing us to discuss whether or not EUID was a mistake or not :)
Ipv6 was a protocol engineered in isolation from the social / political environment it had to be adopted in.
A successor to ipv4 wasnt a technical issue. duh, use longer addresses. The problem was social.
It's a miracle it was used at all
What's annoying about ipv6 discussions is that the ipv6 people are incredibly condescending when the problems of its adoption were engineered by them.
Exactly. IPv6 was developed in the ivory towers where it was still assumed that everyone wanted to be a full participant of the internet.
But the social/political environment was that everyone just wants to be a passive consumer, paying monthly fees to centralized hosts to spoon-feed them content through an algorithm. For that, everyone being stuck behind IPv4 CG-NAT and not being able to host anything without being gatekept by the cloud providers is actually a feature and not a bug.
4 replies →
The problem of IPv6 adoption in the US was largely engineered by major ISPs not caring while hardware manufacturers take their cues from major ISPs.
IPv4 link local addressing is awesome for direct PC to PC connectivity with no hassle
Well you will be happy to hear that ipv6 has the same thing with the FFfe::/10 network just like 169.254.0.0/16 apipa range
1 reply →
So like plugging two laptops together? Honestly curious, I can't recall ever seeing anyone using it and the situations that it seems like it should be good for, like initial configuration of stuff coming out of the box, instead come with instructions for setting specific IPv4 addressing or use DHCP. Possibly a lack of some LLNMR equivalent at the time.
2 replies →
80% of my career knowledge as a devops engineer, systems administrator, and IT engineer has been on the job training. That's just how it works.
The real reason is IT people hate ipv6. They want NAT. They don't want all the security holes and extra complexity. I don't want having to work with a network stack that is poorly supported by some switches and routers.
> Everything he learned about IPv6, he had to learn on his own or on the job.
Replace "IPv6" in that sentence with any practical knowledge or skill and it's probably true for my entire master's degree....
Weird, I graduated from RIT in 2009 with a B.S. in Applied Networking and Systems Administration and we covered IPV6 quite a bit
I certainly can validate this anecdote, I also had to learn almost everything about IPv6 myself.
IPv6 was superceded by NAT a long time ago. It will die a slw and quiet death which is why it is now being ignored by training facilities and experts worldwide.
Oh no, somebody should warn all the ISPs deploying IPv6-native connections with v4 reachable over some fallback technology (464XLAT, DS-Lite, NAT64 etc.) to their hundreds of millions if not billions of customers!
--Sent from my IPv6
The only ISPs issuing IPv6 only connections are mobile device operators and Telcos. THey are a small subset of ISPs in the world and IPv6 only connections will never gain any traction outside of that world.
I agree it will not die so I retract that statement, but it will never fully replace IPv4 in standard wired internet connections.
Digital Ocean didn’t even have an ipv6 address on by default in the droplet I created last week. It’s just a switch to flip, but I’ll bet the support costs of hobbyists/enthusiasts not realizing they needed to also write firewall rules, make sure ports weren’t open for databases and things like that for ipv6.
My memory of IPv6 is getting waves of support tickets from people who took their (already questionable) practice of blocking ICMP on IPv4, blocked ICMPv6, and then got confused when IPv6 stopped working.
1 reply →
It's a "just doesn't work" experience every time that I try it and I don't experience any value from it, it's not like there isn't anything I can connect to on IPv6 that I can't connect to on IPv4.
My ISP has finally mastered providing me with reliable albeit slow DSL. Fiber would change my life, there just isn't any point in asking for IPv6.
Also note those bloated packets are death for many modern applications like VoIP.
11 replies →
Last time I looked at Digital Ocean they had completely missed the purpose of IPv6 and would only assign a droplet a /124 and even then only as a fixed address like they were worried we are going to run out of addresses.
9 replies →
I use IPv6 on my authoritative DNS servers and that's basically it. To your point keeping it disabled on all my hobby crap keeps everything simple for me. If someone can not reach IPv4 then something is broken on their end.
1 reply →
NAT doesn't solve everything, and creates a whole new class of problems that you can just avoid by adopting IPv6 natively. And it's definitely not being ignored at larger companies.
In particular, just off the top of my head...
- T-Mobile US doesn't even assign clients an IPv4 address anymore. Their entire network is IPv6 native.
- Many cloud providers charge extra for IPv4 addresses, but give IPv6 addresses out for free.
For trivial cases NAT is easy, for complex situations it's a nightmare. I've been fighting a lonely battle against multiple-NAT VPNs as being the solution to the wrong problem for longer than I care to remember, and I'm tired boss. A few years ago we had a client site go offline because a local network guy just didn't like IPv6 and turned it off, not realizing that a huge amount of stuff was happening automatically and that's why he hadn't been needing to work on it.
This is not even funny to read, given huge networks like T-Mobile USA being IPv6-only.
Yep, mobile device space ISPs again which is what keeps being argued. IPv6 only connections will never gain full traction outside of the mobile marketplace.
They are using IPv6 as a fancy transport protocol for IPv4 NAT.
10 replies →
It was?
Isn’t it what all the cell phones networks use these days? And most ISP’s?
They may hand the end user device a IPv4 address but don’t they actually use IPv6?
Yes as I said in a sibling post the telcos are the only ones using it, and that is the only reason that graphs like the google client one exist. That is only because it already exists and is cheaper than using NAT when you have hundreds of millions of clients.
IPv6 only ISPs will never leave the mobile space.
8 replies →
AWS charges for ipv4 addresses but ipv6 addresses are free. ipv4 with NAT doesn't supercede ipv6, it just extends its life.
What are you even basing that on? Here are some facts:
- You have to pay money to get a static IPv4 address for cloud machines on eg AWS. Anything needing a static IPv4 will cost more and more as demand increases. NAT doesn’t exactly fix that.
- Mainstream IoT protocols have a hard dependency on IPv6 (eg Matter/Thread). Not to mention plenty of 5g deployments.
- Many modern networks quietly use IPv6 internally. I mean routing is simpler without NAT.
So it almost definitely won’t die. It’s more likely it’ll slowly and quietly continue growing behind the scenes, even if consumers are still seeing IPv4 on their home networks.
IPv4 addresses have been dropping in price for a few years and are cheaper in real terms than at my point in the last 15
2 replies →
Matter/Thread use private IPv6 addresses so it's just an implementation detail. Nobody is exposing light switches to the public Internet.
NAT fixes it in the sense that blocks become available when providers switch to CGNAT.
https://www.google.com/intl/en/ipv6/statistics.html
People love this graph and regularly tout it as if it explains full internet usage. Especially when they dont bother to add any explanation or comment alongside it.
This graph is mainly due to the fact that telcos use IPv6 for mobile devices, nothing more. Over time you will see that graph flatline and peter out as mobile device uage reaches critical mass.
28 replies →
What is the source of the seasonality in that graph? Spikes up a little each summer.
2 replies →