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

More than 1/3 of all access to Google is now over IPv6

google.com

Post reply on HN
Print view View original

141–150 of 237 posts

Re: More than 1/3 of all access to Google is now over IPv6

#141

Post by Dagger2 »

yyyk wrote: post #85
treysis wrote: post #76

Earlier quoted context omitted.

Yes. It offers better scalability! E.g. at my university, our institute has its own IPv4 /24. That means a maximum of ~253 or so devices. We have exhausted that number and for every new client, an old one has to go. Now with a /64 this would be no problem at all. To mitigate this the IT department is currently using VLAN-tagging for some of the devices. But it gets convoluted. And the more convoluted it gets, the hig…

In IPv4-land, it's not an 'internal network' if every device must have an externally routable address - if it is required, IPv6 has an obvious advantage. In the requested scenario (internal Enterprise network), the proper comparison should involved IPv4's 16 million private addresses; I can't see any practical advantage to IPv6 there.

It's an internal network if it's inside your networking perimeter. The type of addresses you use on it are irrelevant to that.

Most enterprise networks need to be routed to other networks -- either other networks inside the enterprise, networks inside other enterprises or the internet. To make that work sanely all of the networks involved need to use address ranges that don't overlap. RFC1918 is pretty much the definition of overlap, and using it makes things way more complicated than they would otherwise be.

Re: More than 1/3 of all access to Google is now over IPv6

#142

Post by Dagger2 »

walton_simons wrote: post #90
mehrdadn wrote: post #86

Earlier quoted context omitted.

> What complexity? 1. What the hell is DHCP-PD and is it better on or off? 2. What are 6to4, 6in4, 6rd, etc. and should the user care? 3. When should autoconf be stateless vs. stateful? I thought the point of IPv6 was to allow things to be stateless? 4. When should DHCPv6 be enabled vs. disabled? Why the hell is this even a question on some routers if devices are supposed to be "autoconfigurable without DHCP"? 5. Wha…

Yeah, this is exactly the sort of thing I'm talking about. Lots of additional overhead and work, and — again, speaking as a home user — no apparent benefits for dealing with it all. I get all the concerns about CGNAT and so on, but that's something for the ISP to figure out. If I get a message one day saying that my connection speed is about to drop 30% because of my insistence on IPv4, I will of course react! The qu…

And ISPs are figuring it out. All you need to do is leave v6 on in the routers they ship out.

But if you're running your own router then you're taking over part of the responsibility, so you need to handle your part of it.

Re: More than 1/3 of all access to Google is now over IPv6

#143

Post by Dagger2 »

BadBadJellyBean wrote: post #29
hansjorg wrote: post #25

China is aiming for 100% in 2025 according to APNIC, so they've got some ways to go from the 0.53% seen by Google. Is the great firewall ipv4 only perhaps?

Google is blocked in China. So I guess not a lot of traffic is coming from Chinese IPs.

That would just lower the number of samples though. For the blocking to drive the v6 percentage down, it would have to happen more often on v6 than on v4.

...which is certainly possible, but just "Google is blocked" alone doesn't cover it.

Re: More than 1/3 of all access to Google is now over IPv6

#144

Post by Dagger2 »

Arnt wrote: post #21
Faaak wrote: post #7

Earlier quoted context omitted.

Nearly all modern operating systems use the IPv6 privacy extension, which creates random addresses periodically. Thus the tracking is randomized to the /64 (which is an end customer). This is more or less the same as with IPv4 (without CGNAT)

Is a /64 really an end customer? How common is that? Both of my ISPs hand out new ISPs hand out new /64 prefixes every so often. One does it every half-hour, the other is comparable but I don't remember. I think they do it to encourage business customers to get more expensive connections, but the effect on my connections is that each device's IP address changes every few minutes, within the ISP's /32, and doesn't rep…

/64 is a subnet. End customers get /56 to /48 so that they can run multiple subnets. It is, unfortunately, more common than it should be for ISPs to not even give the minimum of that range though.

Changing your prefix every half an hour would be... unusual, at least, although again I wouldn't put it past some ISPs to do that (possibly unintentionally)... but sounds like your prefix might be changing every time you do a DHCPv6-PD request, which makes me wonder if your DHCPv6 client might be generating a new DUID for every request rather than remembering it.

Re: More than 1/3 of all access to Google is now over IPv6

#145

Post by amluto »

yyyk wrote: post #85
treysis wrote: post #76

Earlier quoted context omitted.

Yes. It offers better scalability! E.g. at my university, our institute has its own IPv4 /24. That means a maximum of ~253 or so devices. We have exhausted that number and for every new client, an old one has to go. Now with a /64 this would be no problem at all. To mitigate this the IT department is currently using VLAN-tagging for some of the devices. But it gets convoluted. And the more convoluted it gets, the hig…

In IPv4-land, it's not an 'internal network' if every device must have an externally routable address - if it is required, IPv6 has an obvious advantage. In the requested scenario (internal Enterprise network), the proper comparison should involved IPv4's 16 million private addresses; I can't see any practical advantage to IPv6 there.

16 million is way too few because of collisions and the birthday paradox. Even with a tiny corporate network, you are likely to connect to other companies’ networks, and, with private addresses, no one coordinates allocations. Collisions are quite likely.

Also, the major players probably have more than 16 million devices, especially if you could VMs.

Re: More than 1/3 of all access to Google is now over IPv6

#146

Post by yyyk »

Dagger2 wrote: post #141
yyyk wrote: post #85

Earlier quoted context omitted.

In IPv4-land, it's not an 'internal network' if every device must have an externally routable address - if it is required, IPv6 has an obvious advantage. In the requested scenario (internal Enterprise network), the proper comparison should involved IPv4's 16 million private addresses; I can't see any practical advantage to IPv6 there.

It's an internal network if it's inside your networking perimeter. The type of addresses you use on it are irrelevant to that. Most enterprise networks need to be routed to other networks -- either other networks inside the enterprise, networks inside other enterprises or the internet. To make that work sanely all of the networks involved need to use address ranges that don't overlap. RFC1918 is pretty much the defin…

It's fine for IPv6 engineers to want to make everything directly routable, but in IPv4 the difference still matters, and therefore that's the scenario we need to compare against.

The IPv6 enterprise argument is 'You might one day have to renumber some devices because some other enterprise might have set up overlapping addresses, therefore you definitely have to renumber everything now, and oh, every time the ISP decides they don't like you, because private addresses are icky'. Obviously that's a hard sell.

Re: More than 1/3 of all access to Google is now over IPv6

#147

Post by john-shaffer »

dijit wrote: post #40

And yet, when I beg my google cloud rep for IPv6 addresses on instances (or on anything that isn’t the load balancer) I get told that it is not on the immediate roadmap. The cloud providers have pushed back ipv6 adoption so hard imo. At least native ipv6 access. I know they’ve thrown in some token support and you /can/ make something work; but compared to VPS providers which consistently deliver machines with IPv6 ad…

AWS support seems fine to me. You have to do a little setup, but nothing difficult.

Re: More than 1/3 of all access to Google is now over IPv6

#148

Post by dxld »

MayeulC wrote: post #114

I host some services at home, mainly targeted at friends and family. Some are IPv6-only, because it's much easier to manage from my side. I whish I could add an A record for these that pointed to a reserved IP address that would inform clients the service is IPv6-only. For now, I just don't put any, and browsers just display a generic error. Since some DNS don't answer with IPv6 addresses, the browser couldn't even p…

You might be interested in this service: https://no-ipv4-here.ungleich.ch/

Re: More than 1/3 of all access to Google is now over IPv6

#149

Post by tylerl »

dijit wrote: post #40

And yet, when I beg my google cloud rep for IPv6 addresses on instances (or on anything that isn’t the load balancer) I get told that it is not on the immediate roadmap. The cloud providers have pushed back ipv6 adoption so hard imo. At least native ipv6 access. I know they’ve thrown in some token support and you /can/ make something work; but compared to VPS providers which consistently deliver machines with IPv6 ad…

IIRC, all GCPs IPv6 support is complicated by the fact that they adopted IPv6 from the get-go for internal routing, and layer the user-visible virtual address space on top of it, embedding the user-visible addresses inside the invisible "actual" VM addresses, and that layering strategy allows for something super amazing or fast or something. Something like that.

So then you ask the engineers, "when are you going to adopt IPv6?" And they're like: "What do you mean? We've never NOT used IPv6 for everything important."

On the one had my GCP server's "native" IP address that the OS sees is always an IPv4 address. On the other hand, it's always in the 10.x.x.x/8 range. Everything else is NAT and LB.

Re: More than 1/3 of all access to Google is now over IPv6

#150

Post by yyyk »

amluto wrote: post #145
yyyk wrote: post #85

Earlier quoted context omitted.

In IPv4-land, it's not an 'internal network' if every device must have an externally routable address - if it is required, IPv6 has an obvious advantage. In the requested scenario (internal Enterprise network), the proper comparison should involved IPv4's 16 million private addresses; I can't see any practical advantage to IPv6 there.

16 million is way too few because of collisions and the birthday paradox. Even with a tiny corporate network, you are likely to connect to other companies’ networks, and, with private addresses, no one coordinates allocations. Collisions are quite likely. Also, the major players probably have more than 16 million devices, especially if you could VMs.

Every sane large internal network is segmented. I doubt all the "more than 16 million devices" need to talk to every other device. The IPv4 limit is an annoyance, but not a big one for the typical internal networks.
Post reply on HN