India has Highest Rate for IPv6 Adoption, While US Ranks Sixth (aelius.com) 147
Nearly half of the U.S. (47.5%) has migrated to IPv6, ranking the U.S. sixth in percentage of users who have migrated to the latest version of the Internet Protocol. That estimate is based on an analysis of data about Google users — which also finds that India leads the world, with a 61.67% adoption rate. Inside.com's developer newsletter reports:
The figures come from the latest Google IPv6 Statistics data organized and ranked by BBC Radio and Music's Lead Technical Architect for his blog. Malaysia and French Guiana came in second and third, respectively...
IPv6 is considered far more secure, faster, and powerful than its predecessor, IPv4, while offering more IP addresses. Test your IPv6 connectivity here.
After third place there's a close three-way race, where the U.S. has dropped from the #4 position earlier this week. As of today the U.S. is now in 6th place...
Fourth is France (48.38%) and Fifth is Taiwan (48.0%).
Canada ranks #20 (36.59%) and the UK ranks #24 (33.27%).
IPv6 is considered far more secure, faster, and powerful than its predecessor, IPv4, while offering more IP addresses. Test your IPv6 connectivity here.
After third place there's a close three-way race, where the U.S. has dropped from the #4 position earlier this week. As of today the U.S. is now in 6th place...
Fourth is France (48.38%) and Fifth is Taiwan (48.0%).
Canada ranks #20 (36.59%) and the UK ranks #24 (33.27%).
Doesn't tell the whole story (Score:5, Insightful)
Re:Doesn't tell the whole story (Score:4, Informative)
Re:Doesn't tell the whole story (Score:5, Informative)
Re:Doesn't tell the whole story (Score:5, Insightful)
For a lot of people it's just not worth the hassle. IPv4 works, NAT is okay for them, so why change? Larger orgs seem some benefit but even there it's not like things are actually broken at the moment.
Re: (Score:2)
For a lot of people it's just not worth the hassle. IPv4 works, NAT is okay for them, so why change? Larger orgs seem some benefit but even there it's not like things are actually broken at the moment.
The point is maintaining a network that provides the same opportunities for everyone not just lucky rich people who managed to gobble up most of the number resources.
With exhaustion of IP resources assignment is now entirely market based with highest bidder winning and everyone else eating dust.
False (Score:2)
IPv6 is considered far more secure, faster, and powerful than its predecessor, IPv4, while offering more IP addresses.
This statement is not just false it's insane. IPv6 has almost exactly the same security architecture as IPv4 and it objectively requires more bytes of overhead than IPv4 leaving less for user data. The only thing in that sentence which is true is that it offers more IP addresses.
Re: (Score:2)
Here (New Zealand) it's the opposite. I am with the same company for my Home and Mobile internet, Home is dual stack, IPv6 and CG-NAT IPv4, mobile is just CG-NAT IPv4.
Re: (Score:2)
the "NAT" is the key to the startling ongoing use of IPv4. Some environments use the limited IPv4 space to discourage users from having exposed, routable IP addresses, and to compel people's personal devices and system servers to remain behind a locally managed NAT.
Re: (Score:2)
The reason for that is that mobile networks came later, have more users and the users are inherently mobile, so allocating ipv4 to each subscriber was simply impractical. You ended up with cgnat instead which is costly, problematic and reduces performance (although the lack of performance was less of an issue with slower mobile networks).
Fixed line networks in developed countries generally don't use cgnat, are not seeing significant new growth and the incumbent providers already have enough ipv4 to service
Re: (Score:2)
Exactly.
The IPv6 transition for phones happened during 3G. 3G supported both IPv4 and IPv6, with IPv6 preferred.
LTE is entirely IPv6 based. Since it's data only, they mandated that LTE shall use IPv6 exclusively.
5G remains IPv6 only for many reasons, including its intended use for IoT applications and thus needing even more IP addresses.
Carriers use CGNAT to help connect the IPv6 mobile network to IPv4 only internet hosts.
It was easier because carriers used 3G to familiarize themselves with IPv6 equipment,
Faster? (Score:5, Insightful)
IPv6 is considered far more secure, faster, and powerful than its predecessor, IPv4, while offering more IP addresses.
[citation needed]
Re: (Score:2)
IPv6 is considered far more secure, faster, and powerful than its predecessor, IPv4, while offering more IP addresses.
While I disagree with the characterization "far more secure" IPv6 does have significant advantages.
Bypass CGN
Naive brute force scanning of global address space is infeasible
Improves peer to peer communications (Games, voice and video chat, remote access..etc) Enabling capability and enhancing privacy vs routing data channels thru someone else's servers.
SPI inherently more secure than NAT: no packet mangling or exploitable ambiguous ALGs.
IPv6 has potential for better route aggregation (limited by whims of TE
Re: (Score:2)
Massive address space allows for remote networks to cause local broadcast traffic.
IPv6 does not use broadcast traffic.
It uses multicast, and more specifically solicited node multicast for neighbor discovery:
https://en.wikipedia.org/wiki/... [wikipedia.org]
So while theoretically scanning a large sequential address range will generate neighbor discovery traffic, in reality most of it will go nowhere as most of the solicited node multicast addresses won't exist on the switch and will be discarded. So it just becomes a flood of junk traffic like any other DoS attempt.
Less efficiency per packet due to overhead of 128-bit addressing. While this is mostly irrelevant and a rounding error for data transfers the majority of packets transmitted over the Internet are quite a bit less than 100 bytes.
Theoretically, although there are variou
Re: (Score:2)
The RIAA/MPAA/etc love IPv6, because it destroys the legal argument that you cannot identify a person with an IP address. Which is true for IPv4 as one IPv4 can hide a family or a more people behind it.
But you can be sure with IPv6, they'll be getting prefixes from your ISP to contact you to identify a specific address or to preserve the IP address logs so they can identify the PC or device they think is pirating.
And since that sort of notification means you have to preserve the logs, joe home user will get
Re: (Score:2)
[citation needed]
https://datatracker.ietf.org/ [ietf.org] There you go. Unfortunately there's no single document that will answer your question, but if you go through the RFCs you can find many standards that detail precisely what additional IPv6 security features are, how it improves routing efficiency, and what all the additional capabilities are over IPv4. I hope I don't need to provide a citation about how it offers more IP addresses, but just in case the Kahn Academy has some introduction to mathematics videos where you can lear
Re: (Score:2)
the Kahn Academy has some introduction to mathematics videos where you can learn about the powers of 2 as well.
No need to be snarky. :-)
if you go through the RFCs you can find many standards that detail precisely what additional IPv6 security features are,
My comment, as the subject suggests, was about "faster" not "security" -- I'm familiar with this stuff, but not an expert...
Re:Faster? (Score:5, Informative)
Sorry am in a bad mood this morning :-)
The faster part comes from massively increased routing efficiency. IPv4 was never designed to be segmented in the way that it currently is. The idea was that large blocks should represent large networks, and each sub division a smaller network within the larger network meaning that when a packet is routed looking up where it needs to go is supposed to be trivial. The current routing tables have 900k entries already meaning every packet needs to be looked up in a massive table in order to be passed on. IPv6 is supposed to slim that back down as there's ample ways to subdivide the far longer address.
It's like postcodes in a way. If you receive a letter with the postcode starting 75XXX you know it needs to be sent to the Texas post office. That only works if there isn't explicably a 75342 in New York, otherwise you need to look up both numbers to be sure which state post sorting centre to send to.
Sidenote: In 2023 we're going to hit a sort of Y2K style bug. Routing tables are expected to exceed 1,048,576 entries which was incidentally the limit of the table size on a lot of core routers. Network providers are scrambling to upgrade these old ones so they don't hit a cap which could potentially lead to advertised routes not being accepted. Incidentally IPv6 will suffer from this as well as the routing table sizes are only 128kB not 1MB for similar model routers (reads: old ones) and while 1MB should be more than enough for everyone, it turns out 128kB is not likely to be.
That's my admittedly limited understanding of BGP anyway :-)
Re: (Score:2)
Re: (Score:2)
We’ve observed that accessing Facebook can be 10-15 percent faster over IPv6. [fb.com]
His contention was that, on average, IPv6 network connections are established 1.4 times faster [www.sidn.nl]
Google's stats [google.com] have a per-country measurement of the latency hit of v6, and for almost all countries it's either 0ms or negative.
Here's a few.
Re: (Score:2)
Re: Faster? (Score:4, Insightful)
It must be those five levels deep nested NAT gateways that are slowing us down.
I know I may be in the minority, but I like NAT over having my (an) IP address directly exposed/available on the Internet.
Re: (Score:2)
I know I may be in the minority, but I like NAT over having my (an) IP address directly exposed/available on the Internet.
NAT is less secure than SPI.
Re: Faster? (Score:5, Informative)
Uninformed comment by someone that doesn't know what:
1) A stateful firewall is
2) What IPv6 privacy extensions are.
1) A stateful firewall notes output requests, and allows replies back. Random data is dropped. NAT is just stateful with an extra step.
2) IPv6 privacy extensions mean that your device will generate random IP addresses to reach hosts. Your ISP assigns you (at least) a /64, and this is like your IPv4 IP. the upper part of your IPv6 never changes (just like your current NATed IPv4 never changes). The lower bits (2^64 of them JUST FOR YOU instead of 2^32 for the whole internet), you can use anything you want. Your OS will generate a random address for every host it visits.
Re: (Score:2)
> IPv6 privacy extensions mean that your device will generate random IP addresses to reach hosts
So what? That lookup of IP addresses is what DNS reports, and it's why rootkitted laptops broadcast their status back to the control nodes. Those IP addresses are also part of the outbound traffic from your home network, and recordable.
Re: (Score:2)
So what?
So the risk from IPv6 is the same as being assigned a single public IPv4 address or being stuck behind a CGN.
That lookup of IP addresses is what DNS reports, and it's why rootkitted laptops broadcast their status back to the control nodes. Those IP addresses are also part of the outbound traffic from your home network, and recordable.
DNS? Rootkits? Rabbit? Flu shot? Someone talk to me.
Re: (Score:2)
That is one obscure reference.
Re: (Score:2)
It means that "security through obscurity" isn't reliable. NAT has been useful to reduce the exposure of entire networks to outside attackers. It's a commonplace ideal for IPv6 that all devices should be exposed to the Internet, and that everyone should have well-configured firewalls to protect them from outside attack. Sadly, it's rare in practice.
The additional risk of IPv6 is primarily one of implementation: it's touted as enabling access to every internal device from offsite, which is in many ways true.
Re: (Score:2)
It also tends to force people to ask for NAT port forwarding or proxy configurations to allow incoming traffic, which can be very useful.
Carrier-grade NAT "can be very useful" to an ISP's investors as the ISP adds a hefty annual upcharge for the port forwarding.
Re: (Score:2)
The ideal is for each network to have a globally unique prefix on it, taken from a range provided by the ISP, and for access control to be done by a firewall. This isn't "exposed to the internet", and it's not rare in practice -- it's how most people get v6. In a company/campus setup, you'd still need to ask for the firewall to be configured to allow a given inbound connection. NAT would just add extra complication and breakage to the setup.
Re: (Score:2)
It means that "security through obscurity" isn't reliable. NAT has been useful to reduce the exposure of entire networks to outside attackers. It's a commonplace ideal for IPv6 that all devices should be exposed to the Internet, and that everyone should have well-configured firewalls to protect them from outside attack. Sadly, it's rare in practice.
Nobody is being asked to rely on security thru obscurity. They are relying upon SPI which is more secure than NAT. Nobody is suggesting anything be "exposed to the Internet" or suggesting host firewalls should be relied upon for protection. As for rare all consumer routers with IPv6 support come with the same SPI policy as NAT provides enabled out of the box. Every single last one of them. Ditto for IPv6 capable firewalls.
The additional risk of IPv6 is primarily one of implementation: it's touted as enabling access to every internal device from offsite, which is in many ways true. This is often done quite badly, where even a very simple NAT compelled by IPv4 address space limitations would reduce the risks profoundly for a company or a campus.
Again this is neither the argument or the real world implementation at all. The
Re: (Score:3)
How is it a random address? They just have to track you by the "upper part".
Re: (Score:2)
...just like they can with the single routable IP in NAT.
Re: Faster? (Score:4, Interesting)
The Privacy Extensions seem to have been designed to only provide as much protection as NAT did, i.e. not very much if it's a domestic connection with only a few devices on it.
It's a missed opportunity.
Re: (Score:2)
The Privacy Extensions seem to have been designed to only provide as much protection as NAT did, i.e. not very much if it's a domestic connection with only a few devices on it. It's a missed opportunity.
IP is the wrong layer for privacy.
Re: (Score:2)
Uninformed comment by someone that doesn't know what:
1) A stateful firewall is
2) What IPv6 privacy extensions are.
I've always had my home systems behind a business-class firewall/router and researched the IPv6 settings and privacy extensions this evening after posting. I had dual IPv4/6 enabled on my router and systems for a while, but hadn't configured the IPv6 stateless settings and router advertisement completely on the router -- the setup seems more complex than IPv4. I did that this evening and confirmed that the IPv6 privacy extensions were enabled on my Linux and Windows systems. I now have both IPV4/6 connecti
Re: (Score:2)
This stateful firewall you talk about... tell me more about them because they seem to be rare devices these days. If you can open a ssh connection, reboot your firewall/nat device and not have the ssh connection drop every time, that isn't a stateful firewall. Another way to check if a firewall is stateful is to figure out how much data it is using tracking the connections. If it is too low, it can not be stateful.
I haven't seen low cost stateful firewall in a long time. Most firewall just assume outbou
Re: (Score:2)
But that is because connection-tracking is kept in RAM. iptables can even sync conntrack between routers so you can have failover between routers. I don't know if it can persist to disk, but since conntracking is so intensive, it doesn't really seem t
Re: (Score:2)
I know I may be in the minority, but I like NAT over having my (an) IP address directly exposed/available on the Internet.
Nothing about IPv6 means you need to expose your IP addresses on the internet. You can still NAT if you feel the need to fuck up end-to-end connectivity and further ensure that we are wholly dependent on cloud services for everything going forward.
Knowing an internal IP is useless. You can still prevent them from doing fuck all with SPI.
Re: (Score:2)
I know I may be in the minority, but I like NAT over having my (an) IP address directly exposed/available on the Internet.
Plenty of correct responses saying how wrong this statement is, but I would add that if you're suitably insane, NAT66 [wikipedia.org] is a thing.
Re: (Score:2)
Some people who run a home server prefer putting the server in a DMZ over A. having to pay a VPN provider every month or year for relay service because the server is behind ISP-managed NAT or B. having to pay a VPS provider every month or year for hosting.
Re: (Score:2)
Re: (Score:3)
I know I may be in the minority, but I like NAT over having my (an) IP address directly exposed/available on the Internet
Don't confuse NAT with a Firewall. Firewalls provide security. NAT is not a security feature - it is a side-effect of IPv4. An IPv6 router can provide all the same protections we have today with IPv4.
Overall, we don't suddenly want every Windows machine on earth to suddenly have TCP port 445 exposed to the internet. Any sane firewall/router ought to prevent that out-of-the-box, regardless of NAT.
Re: (Score:2)
Don't confuse NAT with a Firewall. Firewalls provide security. NAT is not a security feature - it is a side-effect of IPv4. An IPv6 router can provide all the same protections we have today with IPv4.
Thanks. My current router/firewall is a D-Link DSR-250 [dlink.com] and supports IPV4 or dual IPV4/IPv6 -- I have the dual option selected (also not sure if or how well everything on my local net supports IPv6 -- like my TiVO). Until recently, I wasn't getting IPv6 through the router because I hadn't completely configured the stateless and router advertisement settings on the router, but did that the other day (and confirmed that the privacy extensions are enabled on my Linux/Windows systems) -- IPv6 is more complicat
Re: Faster? (Score:4, Interesting)
When the move to classless routing in IPv4 came about, it made better use of IPs, however, it meant lookups of class C, and then individual IP to locate host. So short answer is, IPv6 routing is much faster due to classed routing.
Maybe I'm misinterpreting what you are saying, but IPv6 routing has always been classless. Just like IPv4 has /15, /16, /17, etc., the current IPv6 route table has (looking quickly at a few prefixes on an internet router) /32, /33, /35, /41, /47, /48, and others. To quote from Geoff Huston [apnic.net]:
Routing advertisements of /48s are the most prevalent prefix size in the IPv6 routing table (some 48% of all prefixes), and 75% of the table entries are composed of /48, /32, /44, and /40 prefixes.
Re: (Score:3)
However, in IPv6, we have address blocks of varying sizes that are then allocated to a nation.
In many ways, those are simply classes.
Most of CHina's are
Interestingly, US has smaller blocks, but against, most of ours are in the 26XX: range. [imtbs-tsp.eu]
Brazil and other Latin American nations are in 28XX [imtbs-tsp.eu]
You will find over and over that RIR is headin
Re: (Score:2)
You will find over and over that RIR is heading towards trying to keep nations together in a ip range, similar to how class routing used to work in IPv4. This makes for much faster look-ups on systems.
Registries allocate on a geographical basis, but if a country has two major ISPs, and they buy transit from different Tier 1 providers, then their prefixes will be routed differently on most backbone routers. Even if they make the exact same choice of provider, a backbone router just sees a bunch of /32s or whatever, and has no optimization to make look-ups faster.
Re: (Score:2)
Older equipment didn't need 1...10...0 netmasks. It still works on some software and hardware. It is an very ugly hack but sometimes can be useful for renumbering networks
Class A, B & C went away long before most networking people I know ever set up a network yet it is still in modern documentation and is very well known. Because of that, in my IPv6 class, I use /32 =class A, /64=class C and /48 to /56 as class B. It seems to help explain why you don't want to use a /120 in a v6 network.
How many IPv4 the USA have vs India? (Score:3, Insightful)
Re:How many IPv4 the USA have vs India? (Score:4, Insightful)
That is why I would love to see just 1 large ISP do nothing but IPv6. I had hopes that Starlink would do this.
Re: (Score:2)
History yes, but this history has caused significant inequality today which is very harmful to developing countries. This inequality needs to be fixed, and IPv6 is the fix.
Remember IPv4 was an experimental protocol intended for the ARPANET, it was never intended for a global network. In a speech given a few years ago by Vint Cerf, he stated IPv4 is experimental and IPv6 is the production version.
So what? (Score:2)
India doesn't have SpaceX.
Slow Progress (Score:2)
I have a data center in Toronto where IPv6 isn't available. The world hasn't ended so far.
Re: (Score:2)
I have a data center in Toronto where IPv6 isn't available. The world hasn't ended so far.
Perhaps not for you but a /16 is now priced at a million or more at auction.
Re: (Score:3)
It really doesn't. Lack of address space causes a ton of issues, and mitigating those issues costs us a ton of time and money. Things haven't worked "just fine" for years.
Re: (Score:2)
No, you are, since ipv4 will be very useful and profitable for more than two decades at least.
Should outlaw IPv4 as far as I am concerned (Score:2)
I have a couple of servers living in a web co-host facility that has very good IPv6 support and is close to high traffic hubs. And for some cosmic error of advancement that will never be explained, Comcast has really decent support for IPv6 for its residential cable subscribers.
End result is I use IPv6 a lot and it turns to be a lot more convenient than IPv4 ever was. No NAT, no problem with address space allocation (I have 14 IPv4 addresses that the co-host gave me), no having to re-use addresses all
Re: Should outlaw IPv4 as far as I am concerned (Score:2)
Re: (Score:2)
There are many ISPs which already provide IPv6, but they also always provide some form of backwards compatibility mechanism too - either dual stack, NAT64, DS-Lite etc.
IPv6 works better, and achieves better performance (sometimes significantly so) on these providers, but the users aren't aware why so they don't demand IPv6 from other ISPs. LinkedIn did a study a few years ago and determined that IPv6 was usually quicker, sometimes up to 40% (likely caused by cgnat):
https://www.linkedin.com/pulse... [linkedin.com]
I've had
Reddit, Twitter, Amazon, Twitch, eBay, GitHub (Score:2)
Any site that doesn't enable IPv6 these days is giving a second class experience to many users, and likely drives users away to other "faster" sites.
What would it take to convince users to actively avoid websites that don't support IPv6? The website Why No IPv6? [whynoipv6.com] lists Reddit, Twitter, Amazon, Twitch, Pornhub, XVideos, Yahoo! Japan, eBay, GitHub, PayPal, and Stack Exchange as among the most popular laggards outside China. A lot of these are communication platforms that don't have significant "faster" competition because network effect. I'm guessing many have hesitated because adding IPv6 would amount to amnesty on IP range bans issued for serious violati
Re: (Score:2)
Cute. Of course they can, broadcast+multicast scanning exists on IPv6 just as it does on IPv4. As Cisco Talos proved there are plenty of misconfigured dual-home hosts out their advertising their IPv6 addresses in response to IPv4 UPnP requests, and there will be many other likewise-stupid protocols.
Still remember the day I configured remote access IPv6 only. The complete and total lack of continuous bullshit and resulting log spam is nothing short of amazing.
People can split hairs and nit pick all the ways someone could still scan IPv6 address space (except for "broadcast scanning".. this isn't a thing anymore) till their blue in the face. The fact of the matter is its way way way way way way better with IPv6. Most importantly you and your 64+ bits have a say in how good the difference is by your o
Probably the biggest reason for V6 adoption (Score:2)
Is the large number of call centers across India.
And of course the lack of available V4 addresses in useful chunks.
Re: (Score:3)
Need a large ISP to switch to IPv6 (Score:2)
Re: (Score:2)
A large ISP would probably then try to charge you for each IP address you use, because they're shitty shits who do shitty things.
Re: (Score:2)
I've never heard of any ISP charging customers for individual v6 addresses. I am aware of lots of ISPs not doing that. I'm sure you could find one somewhere if you looked (probably a server host or something similar...), but it's certainly very uncommon.
ISPs that provide less than a /56 are more common, but even then it seems to be less a matter of charging people and more a matter of just failing to provide a minimum service. "ISPs will charge you for each individual IP" is scaremongering rather than somet
Re: (Score:2)
I only get a /64 from my ISP, which means only one vlan can be ipv6. I have six primary vlans at home. Spectrum won’t provide /56’s for residential.
Re: Need a large ISP to switch to IPv6 (Score:2)
Reverse DNS (Score:2)
In my experience, wide configuration of "proper" rDNS for IPv6 addresses in use seems lacking.This makes some e-mail checks (forward/reverse/IP/HELO) not work, and that's my primary area of interest/concern, as such checks are, from my perspective, a must when it comes to combatting spam these days.
i see this issue as an impedement to full, IPv6-only transition for many e-mail server operators.
Re: (Score:2)
> In my experience, wide configuration of "proper" DNS seems lacking
I hope you don't mind that I fixed that statement for you?
Re: (Score:2)
Reverse DNS is only used for spam checking of SMTP connections, there is virtually nothing else that uses it.
If you are hosting a mail server, you setup RDNS, if you're doing anything else then you probably won't bother. Where there are stray PTR records on legacy IP that is often caused by address reuse from the days when more people used IRC, or when people actually bothered to set RDNS. Nowadays virtually noone does.
There's nothing inherent in RDNS either, it's just assumed that anyone configuring a mail
you don't say (Score:2)
The nation with the most IPv4 space isn't ready to trade in their valuable addresses?
Maybe enterprise routers will pony up and pay for licenses for the firmware updates that include IPv6. I keep running into colos that still don't have good native IPv6 support, which trickles down into VPS services that force you to have an IPv4 address.
I'm sure in 10 years we'll have completed the transition ... mostly.
Re: (Score:2)
All equipment made for many years is IPv6 capable by default. If a provider is running equipment which is not IPv6 capable then this is a VERY BAD SIGN. This equipment is likely to be hugely out of date, probably entirely end of life and unsupported by the vendor at this point which means it's likely to have very serious security holes.
A lack of IPv6 support is an indicator of SERIOUS problems at a provider, the security problems mentioned above plus the general lack of support for modern protocols. I would
Re: (Score:2)
If a provider is running equipment which is not IPv6 capable then this is a VERY BAD SIGN.
I can get you a list of colos that are used for enterprise that have this problem. I wish it was just a distant memory, but I setup a system in June this year and was shocked when I had IPv6 problems. In the end I had to setup a tunnel broker. A very bad sign in deed, but typical.
This equipment is likely to be hugely out of date
What happens is vendors provide a support branch where you get security updates for an old version of their OS. I know Juniper and Cisco did this, I suspect other vendors as well. If the new OS supports old hardware, you usually ha
Re: (Score:2)
Cisco support old versions for years with security updates, but at some point those updates stop and the device becomes end of life so you can't even get paid support for it anymore. Devices which lack IPv6 have long since fallen into the EOL category, multiple generations of hardware with IPv6 support are also EOL too. If you have something which doesn't support IPv6 from a major vendor like Cisco then it is way out of support and has long since passed the point of having any security updates.
Re: (Score:2)
Sorry, but that's not really true. It depends on what is meant by IPv6 support. Yes, the equipment can switch IPv6 and with a little effort can tunnel them. But without configuration in RIP a proper route can't be established, so your IPv6 networks end up being isolated.
I will admit, it's been a while since I worked for Cisco, but I'd be surprised if that stodgy company's policies have changed at all.
NOBODY has migrated to IPv6 (Score:2)
Re: (Score:2)
That's kinda poorly informed. There's a significant number of sites out there with v6 (not as many as there should be, mind), and if you have v6 a significant amount of your traffic (more than half by byte count, on average) will flow over v6.
The point of v6 isn't to let you do new things, but to allow you to continue doing things you used to be able to do on v4, but can't due to address space exhaustion. Sometimes (but not always) there are workarounds for v4, but they're costly and a pain in the neck. Sim
Re: (Score:2)
Re: (Score:2)
CGNAT should be punishable by death.
Lots of dead people then in ten years.
Re: (Score:2)
There is nothing you can do with IPv6 that you can't do with IPv4
Except accept incoming connections in countries where all ISPs lack enough IPv4 space for all subscribers. This means you can't run a personal website, media server, or game server from home without paying a significant amount of money per month for a VPN to relay requests between the Internet and your home server behind carrier-grade NAT.
Re: (Score:2)
Re: (Score:2)
Except accept incoming connections in countries where all ISPs lack enough IPv4 space for all subscribers. This means you can't run a personal website, media server, or game server from home without paying a significant amount of money per month for a VPN to relay requests between the Internet and your home server behind carrier-grade NAT.
You explained yourself why IPv6 adoption rate of ISPs are so slow. Why take time and effort such that customers no longer need to pay premium when they want to host public server at home / office? This is anti-profit.
Something's wrong at Google IPv6 stats (Score:2)
What happened to the all-time leader Belgium ?
Looks like there's something wrong with the Google IPv6 stats : https://www.aelius.com/njh/goo... [aelius.com]
Re: (Score:2)
My belgian major ISP (proximus) doesn't provide native ipv6 on my (business grade) connection.
Re: (Score:2)
Telenet and several others have been for many years. Belgium was number 1 for a long time.
Re: (Score:2)
Have you reported this lack of IPv6 service to your ISP? If so, what was the reply? And what other business-class ISPs are available at your address?
Re: (Score:2)
Yes, answer is: "We do not support ipv6 on your line". I guess they need to migrate most "older" business customers to newer backbones and that would probably involve losing/changing my fixed ipv4.
I can't really change my ISP if I want to keep my fixed IPv4s.
Re: (Score:2)
What happened to the all-time leader Belgium ? Looks like there's something wrong with the Google IPv6 stats : https://www.aelius.com/njh/goo... [aelius.com]
Something odd happened between 2021-08-11 (49.2%) and 2021-08-18 (17.85%) so it does look like there is a measuring problem this week. They should still be ahead of the US.
Google should probably fix Android then (Score:2)
https://issuetracker.google.co... [google.com]
Re: (Score:2)
No one cares what you do internally. And dual stack means you can have hosts speaking whatever you want, as long as any public-facing servers have IPv6.
Again, no one cares what you do internally.
IPv6 support in Windows is available since 2001 in
Re: (Score:2)
Virtually all cloud infrastructure runs internally on IPv4. All corporate LANs are IPv4.
There is nothing wrong with this. People should feel free to use IPv4 forever locally so long as Internet traffic is IPv6.
All products are designed for IPv4, because the backend systems are IPv4, the developers' desktops are IPv4, their ISPs are IPv4, the services they use on a daily basis are IPv4. The routers and switches are IPv4. Every guide on the internet for doing anything involving an IP address is for IPv4.
Yes of course, what's important is they are also capable of IPv6.
The only thing that will drive IPv6 adoption is if the government starts forcing people to use it by default. They can start by adding the requirement to all their contracts with vendors, service providers, contractors.
https://www.ferc.gov/internet-... [ferc.gov]
And then they can start introducing rules that all *new* network products sold have to ship with IPv6 support out of the box.
What isn't shipping with IPv6 support out of the box?
Re: (Score:2)
> People should feel free to use IPv4 forever locally
This is the reason IPv6 doesn't have larger adoption. People are lazy. If they only ever use IPv4 internally, they'll only ever use IPv4 externally. And there are additional problems with bridging IPv4 and IPv6 that we need to avoid.
Why should people feel free to use IPv4 forever? We have outgrown it, and it needs to die so we can drive adoption of IPv6. As long as IPv4 is an option, people will prefer it to IPv6, because IPv6 is more of a pain in the
Re: (Score:2)
This is the reason IPv6 doesn't have larger adoption. People are lazy. If they only ever use IPv4 internally, they'll only ever use IPv4 externally. And there are additional problems with bridging IPv4 and IPv6 that we need to avoid.
You don't have to bridge anything if you don't want to.
Why should people feel free to use IPv4 forever?
There is no reason to get rid of it.
We have outgrown it, and it needs to die so we can drive adoption of IPv6. As long as IPv4 is an option, people will prefer it to IPv6, because IPv6 is more of a pain in the ass, because it is new and weird.
Ends don't justify means.
Cloud services.
What does cloud services mean? Cloudflare, Akamai, AWS and Azure all support IPv6.
Most are IPv4-only. And most software is only set up for IPv4
Nonsense, ditto for software.
Re: (Score:2)
Setting AWS up to use IPv6 is a pain in the ass.
I want to do it at work, but it would require basically rebuilding the entire subnet structure, which means a LOT of paperwork to remain compliant since we need to touch basically everything.
Re: (Score:2)
Tech companies corporate LANs are often IPv6, read the posts by Microsoft and Facebook a few years back now.
The US government is already forcing IPv6, as is China. Several governments such as Israel, India, etc are also requiring that ISPs provide IPv6 by default.
Re: (Score:3)
IPv6 is not a security nightmare, ignorance of IPv6 is a nightmare.
If you configure IPv6 properly, it is better than legacy IP in many ways, and isn't worse in any way.
You don't give away internal addresses even in the default configuration, because client hosts use ephemeral privacy addresses which rotate.
Legacy IP is also meant to be configured with routable addresses on both sides of the firewall, the only reason this isn't typically done is because the cost of doing so outweighs the drawbacks of NAT.
IPv
Re: (Score:2)
Yes, some of us having been running production IPv6 networks for many years, and were involved in the 6bone trials more than 20 years ago.
From my own experience i know that IPv4 causes a lot of problems, and requires all manner of kludges to work around them. Those who have never used IPv6 properly simply aren't aware how much hassle is avoided, they are the downtrodden who have come to accept the kludges as normal and can no longer imagine the existence of a better way.
Even if they do implement IPv6, they
Re: (Score:2)
If you look at Facebook's [facebook.com] stats for 2018-2021, Belgium has gone from 50% to 55%, the United States from 47% to 60%, and India from 47% to 69%. It looks like Belgium has been relatively static, and other countries have had a growth spurt in the last few years.
Google reports 47.5% for the US, and Facebook reports 60.24%, so there is some difference in the types of user and the populations being measured.
Re: (Score:2)
Re: (Score:2)
OPNsense.