Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

40% of Google users now connect via IPv6

google.com

Post reply on HN
Print view View original

81–90 of 353 posts

Re: 40% of Google users now connect via IPv6

#81

Post by aoetalks »

ignoramous wrote: post #56
TekMol wrote: post #41

My website has a few hundred thousand users. My development is completely user driven. I get hundreds of feature requests and other emails regarding my site per month. So far, not a single user mentioned IPv6. So I never added an AAAA record. The codebase has grown for over 10 years now. God knows what subtle breakages would happen when IPv6 requests come in. I wonder if I ever will add IPv6 or just leave the site ru…

> So I never added an AAAA record. IPv4 servers are reachable by IPv6-only clients. ipv6test.google.com famously has only an A (IPv4) record for a reason. > God knows what subtle breakages would happen when IPv6 requests come in. Unless your product deals with the nitty gritty of the networking stack itself, pretty sure everything continues to hum along just fine, if the basics are covered.

>IPv4 servers are reachable by IPv6-only clients

How does this work?

Re: 40% of Google users now connect via IPv6

#82

Post by imtringued »

p1mrx wrote: post #49
sneak wrote: post #44

Earlier quoted context omitted.

Get a raspberry pi, set up two vlans, run wireguard, and send 100% of your uplink traffic to a remote vpn endpoint. This gets you off their IP, and also has the added benefit of not letting them analyze your traffic.

Having a non-residential IP address is likely to get you blocked from services like Netflix. It would be nice if there were some way to decouple connectivity from addressing, without becoming a second-class citizen of the internet.

I don't use Netflix but that never happened to me.

Re: 40% of Google users now connect via IPv6

#83

Post by neilalexander »

ajsnigrutin wrote: post #74
neilalexander wrote: post #57

Earlier quoted context omitted.

> Firewalls are more complex on IPv6 This is not true. Firewall rules work exactly the same in IPv6 land as they do in IPv4 land. > you need to pass a bunch of ICMPv6 through, to make it work Indeed you shouldn't block ICMPv6, but that is not really making anything "more complex". > some residential routers have very bad or even zero firewall support for ipv6 Is there a proven set of routers that go through the troub…

> Firewall rules work exactly the same in IPv6 land as they do in IPv4 land. Yes, rules do work exactly the same, but with IPv4, you just let all the connections out through, and let just the established and connected ones back. > Indeed you shouldn't block ICMPv6, but that is not really making anything "more complex". But it is... you need a bunch of new rules to pass through, limit or block a bunch of ICMPv6 messag…

> Yes, rules do work exactly the same, but with IPv4, you just let all the connections out through, and let just the established and connected ones back.

The same is typically true of IPv6 for default configurations. You aren’t required to allow IPv6 hosts to accept unsolicited incoming traffic.

> But it is... you need a bunch of new rules to pass through, limit or block a bunch of ICMPv6 messages.. there's a whole RFC just for that - https://datatracker.ietf.org/doc/html/rfc4890

With the exception of home agent, mobility and other IPv6-specific messages, many of these recommendations also hold true for IPv4. It’s just that nobody really bothers to think that deeply about it, block all ICMP and then are shocked_pikachu_face when Path MTU discovery etc don’t work.

Re: 40% of Google users now connect via IPv6

#84

Post by neilalexander »

fulafel wrote: post #78
neilalexander wrote: post #53

Earlier quoted context omitted.

Why should an unstable DUID matter? The prefix won't change.

What besides presenting the same DUID would cause you to get the same prefix again?

The host portion of the address changes. The prefix doesn’t.

Re: 40% of Google users now connect via IPv6

#85

Post by Unklejoe »

fulafel wrote: post #78
neilalexander wrote: post #53

Earlier quoted context omitted.

Why should an unstable DUID matter? The prefix won't change.

What besides presenting the same DUID would cause you to get the same prefix again?

Could it be based on whatever physical port you’re attached to on the ISP’s network?

Re: 40% of Google users now connect via IPv6

#86

Post by tyingq »

vmoore wrote: post #2

Anyone go out of their way to disable IPV6 on all their systems, even at the hardware level, like on routers etc? What's a good reason to avoid IPV6?

Yes, there are situations like "The crap VPN (hello AnyConnect) my work makes me use doesn't work if IPV6 is enabled. And I could troubleshoot it, but it's easier to disable IPV6 on my PC".

Re: 40% of Google users now connect via IPv6

#87

Post by BenjiWiebe »

stereo wrote: post #62
NelsonMinar wrote: post #48

Earlier quoted context omitted.

And adds 20ms+ latency, another complex point of failure, and potential problems depending on the reputation of the IP address block of wherever you're hosting the endpoint. I've used VPNs to smooth over various Starlink problems since I got the service and it helps but it's not a great solution.

My he.net ipv6 tunnel adds about 2ms of latency on average, which I can totally live with.

Can you use that over a CGNAT?

Re: 40% of Google users now connect via IPv6

#88

Post by zinekeller »

aoetalks wrote: post #81
ignoramous wrote: post #56

Earlier quoted context omitted.

> So I never added an AAAA record. IPv4 servers are reachable by IPv6-only clients. ipv6test.google.com famously has only an A (IPv4) record for a reason. > God knows what subtle breakages would happen when IPv6 requests come in. Unless your product deals with the nitty gritty of the networking stack itself, pretty sure everything continues to hum along just fine, if the basics are covered.

>IPv4 servers are reachable by IPv6-only clients How does this work?

It unfortunately involves CGNATted IPv4, but it's either DS-Lite (https://datatracker.ietf.org/doc/html/rfc6333) where the end-user router emulates a dual-stack network but encapsulates IPv4 traffic on IPv6 from the perimeter to the CGNAT device at the ISP or just plain NAT64/DNS64 (https://datatracker.ietf.org/doc/html/rfc6147) where IPv4 traffic is relayed at dedicated IPv6 addresses operated by the ISP with the help of special DNS (which might be the GP meant for IPv6-only networks, but it tends to be unreliable for a lot of reasons).

In practice DNS64 is now being removed from a majority of networks because some specialty applications (I say "specialty" but these are work VPNs, conference systems and the like) reacts badly (because usually they can't understand IPv6 in the first place), replaced by either DS-Lite or plain dual-stack (possibly with CGNAT for IPv4).

Re: 40% of Google users now connect via IPv6

#90

Post by skrause »

michaelbuckbee wrote: post #13

I'm curious how much of this is mobile vs desktop. My assumption is that a much larger percentage of the traffic on mobile is ipv6.

In Germany, where the Google statistics show 64% IPv6 adoption, mobile carriers were actually the last to support IPv6, but even the last mobile carrier enabled it around a year ago.

Now it's mostly businesses that are still not using IPv6.

Post reply on HN