There is a lot of NAT going on these days, and a lot of shared addresses. All those devices aren't all online at the same time.

As I see it, IPv6 can't come fast enough. NAT really needs to die a death so people can actually use the internet fully, not only as a client or using hack-y work-arounds.

I want ipv6 as much as anyone, but we are not spending enough time preparing for NATs death. It's responsible for far more security than we as a collective would like to admit and there's a lot more we could be doing now to get ready.

That's a common misconception. NAT isn't a security feature but rather a feature of a stateful firewall, which is. There's no reason to remove the firewalls that are in place now when ipv6 happens.

If firewalls are even needed, a recent poll on an IPv6 professional forum ended with 50/50 split between opt-in and opt-out for IPv6 firewalls in routers of consumer ISPs...

That's surprising and quite concerning. Imagine all the insecure IoT devices running ancient software having a direct connection to the Internet... It would be even more concerning if they were shipping routers without any firewall functionality at all. NAT basically requires a firewall. I hope the thinking isn't if you can do away with NAT you can do away with the firewall.

The thinking of the opt-inners seems to be (roughly) that :

-IPv6 is fundamentally much more secure than IPv4 (no scanning, etc.)

-opt-out is bad for innovation, especially since the cheap default ISP router firewall software is likely to not even allow opt-out for any other protocols than TCP and UDP. (Heck, these days on IPv4 even anything different than HTTPS can be problematic...)

-reliance on router firewalls is bad because they incentivize sloppy device security - the manufacturers should be instead liable when they are at fault for screwing it up (also, how many of these "insecure IoT devices running ancient software" are even able to run IPv6 ?)

source : https://lafibre.info/ipv6/ipv6-le-firewall/msg704095/#msg704... (fr)

Incidentally, one of the "big 4" French ISPs "Free" didn't even have an IPv6 firewall on its customers routers between 2008 and 2019, and it's probably still opt-in : 4 months ago : https://fr.answers.yahoo.com/question/index?qid=202008121107... (fr)

So I guess that we're going to see in practice the problems that having no IPv6 firewall causes (most customers not having any idea about what even is a firewall) as it gets more popular... and since Free this summer boasted about reaching 99% IPv6 coverage, and is enabled by default, and can NOT be disabled...

> IPv6 is fundamentally much more secure than IPv4 (no scanning, etc.)

The same was true for ipv4 until about a decade ago.

> opt-out is bad for innovation, especially since the cheap default ISP router firewall software is likely to not even allow opt-out for any other protocols than TCP and UDP. (Heck, these days on IPv4 even anything different than HTTPS can be problematic...)

I can't wait for conficker6 to innovate it's way around the ipv6 net.

> reliance on router firewalls is bad because they incentivize sloppy device security - the manufacturers should be instead liable when they are at fault for screwing it up (also, how many of these "insecure IoT devices running ancient software" are even able to run IPv6 ?)

Sounds like an excellent reason for an opt-out by standard. 99% of the world's internet users wouldn't have a clue how to manage a firewall. Directly connecting all their devices to the internet is an awful idea for 99% of the world.

Your 50/50 example is hugely biased, first it's on a Telco discussion forum so that clearly selects for technical users, then it's on ipv6 which is going to further select for technical people.

Go canvas 100 random people outside a supermarket if they want to have to manually manage a firewall for every device they connect to their network. If they don't give you a blank stare at that question remind them that includes everything from lightbulbs, washing machines, "smart" speakers, to their computers/phones (likely the only thing they think of as being connected to the internet). If you find more than 1 I'll eat my hat.

I don't own a hat.

> Your 50/50 example is hugely biased, first it's on a Telco discussion forum so that clearly selects for technical users, then it's on ipv6 which is going to further select for technical people.

As you can see I'm aware of that, they are also aware of that, and the discussion is not so much about themselves (since they know how to configure a firewall or even to install their own router), but about what your "average grandma" should get.

If only average grandma's were just limited to grandma's. I don't know a single person who isn't a gamer or IT person that can properly use a firewall as they exist now.

The overwhelming majority of users doesn't bother in changing any settings, as can be seen from the dramatic changes in IPv6 adoption when an ISP goes from opt-in IPv6 to forced IPv6.

You probably want to read this: https://www.internetsociety.org/wp-content/uploads/2019/02/D...

Thanks. So yeah, it looks like IPv6 is more secure than IPv4... as long as we're talking about competent engineering ! Hopefully this is the case for major ISPs and OSes...

Especially interesting is this RFC : https://www.rfc-editor.org/rfc/rfc6092.html "Recommended Simple Security Capabilities in Customer Premises Equipment (CPE) for Providing Residential IPv6 Internet Service"

It shows that there are lots of different filterings involved, so it looks like that these millions of residential users connected to the IPv6 Internet without router firewalls might still have some router filtering going on ?

Also, it confirms that "The IPv6 stateful filtering behavior described in this document is intended to be similar in function to the filtering behavior of commonly used IPv4/NAT gateways, which have been widely sold as a security tool for residential and small-office/home-office networks.

As noted in the Security Considerations section of [RFC2993], the true impact of these tools may be a reduction in security. It may be generally assumed that the impacts discussed in that document related to filtering (and not translation) are to be expected with the simple IPv6 security mechanisms described here.

In particular, it is worth noting that stateful filters create the illusion of a security barrier, but without the managed intent of a firewall. Appropriate security mechanisms implemented in the end nodes, in conjunction with the [RFC4864] local network protection methods, function without reliance on network layer hacks and transport filters that may change over time. Also, defined security barriers assume that threats originate in the exterior, which may lead to practices that result in applications being fully exposed to interior attack and which therefore make breaches much easier."

So now I'm kind of confused as for the different meanings of 'filtering' and 'firewall' that might be used... The RFC seems to use 'firewall' in the sense of 'customizable firewall', while ISPs still often don't provide other options on their IPv6 'firewall' than 'ON/OFF'...

I'm well aware it's not a security feature and I know there are ways to punch holes, but in practice, a lot of machines are still relying on it. The number of IoT devices alone that would be screwed if they were public is massive.

Yes, everyone should have a hardware firewall, but we both know most people just buy the cheapest thing, and by bad large, real firewall features are mostly targeted toward higher end devices.

20+ years isn't enough?

More seriously; for 99% of people their ISP router handles NAT and firewall duties. Adding DENY ALL inbound and ALLOW ALL outbound isn't a great stretch for them on ipv6.

Of course it's more than enough, but it's still barely done. We should have co2 emissions under control too, but we don't.

Most routers I have seen do not leave ipv6 unprotected. By default they block incoming connections like nat would.

I think we'll still see some people doing NAT66, unfortunately.

Sadly this means that the poorer countries give the same IP to multiple users. I’ve received a “you’re doing that too much” error after the first google search of the day on my phone in South East Asia. I don’t remember which country

That's not exclusive to poorer countries. Cellular carriers in first world countries aren't assigning a public IPv4 address to every cell phone.

I've never really considered this before but it's interesting. Does this happen randomly or is there some geographic/tower based IP sharing?

My guess is that providers allow higher rate limits on developed country ISP IPs for this reason.

I’ve had that happen in Europe so I don’t think it’s just poor countries doing that.

My buddy living in the country side here in Norway switched ISP and ended up behind carrier-grade NAT.

Speed is so much better with new ISP though so he just set up Wireguard to the VPS server he rents to get "his own" IP.

If you believe the Android screen Settings > About > Status, my phone is assigned only an IPv6 address, no IPv4.

Despite this, it's able to connect to IPv4 web servers just fine.

All connections from the IPv6-only phone to IPv4-only web are automatically NAT'd to IPv4 by the cell service provider. I've tested this recently and it uses a different ephemeral source IPv4 after a few minutes when doing this. Tested with HTTP, HTTPS and ICMP ECHO. It is definitely NAT.

At the same time, my connections from the phone to IPv6-only web are not using NAT. The server sees the same source IPv6 as the phone reports as its own.

When I enable tethering on my phone, it creates a local IPv4 wireless LAN. Devices on that LAN such as my laptop access the web using IPv4, which is NAT'd twice: Once on the phone when crossing from the WLAN to the cell network, then by the cell service provider to get an ephemeral source IPv4. This is double NAT.

When the Linux VM on my laptop connects to an internet service and I'm using the Wifi hotspot on my phone, there's yet another NAT in the way, on the laptop itself. This is triple NAT.

All this NAT means it doesn't matter so much that we are out of IPv4 addresses for phones. They can connect to both IPv4-only and IPv6-only services while assigned only a public IPv6.

In fact phones don't need a public IPv6 either. They don't need any public address.

Those NAT'd IPv4 connections don't go over the IPv6 link. They are not being translated to IPv6 and back. Rather, they go over what is effectively a private IPv4 tunnel to the cell provider. Just as IPv4 connections can work like that, so could IPv6 so there's no real need for the phone to report that it has any public IPv6 or IPv4 address at all.

However, mine is currently reporting a public IPv6 and no IPv4, while able to make connections to both.

We are out of ipv4 addresses. That is why amazons stockpile is worth $2B, that's how we are reduced to trading the existing addresses at ever increasing prices instead of getting new fresh ones essentially for free

When i read the title, I first thought to myself that those 2 billion dollars will be pretty much worthless once everyone uses ipv6. But as soon as more ipv6 adresses are used (or CG-NAT), ipv4 adresses will become cheaper, so there seems to be a natural balance which will keep the state of "There are IPv4 adresses left" and "IPv4 adresses will run out sometime soon"

For a while, but eventually we will hit a tipping point where nobody cares about IPv4 anymore.

Well, at some point the governments will actually start doing their jobs and ban IPv4 compatible hardware for sale, just like they did for obsolete analog/digital TV tuners and incandescent light bulbs.

What are you talking about?

IPv4-compatible equipment is bad because it slows down the transition from IPv4 (aka ARPANET) to IPv6 (aka Internet). Since the free market seems to have trouble to find incentives to do it itself, the governments should force it, in the same way that they forced the transitions from obsolete TV tuners and incandescent light bulbs.

ISPs in China is almost exclusively NAT now, you have to specifically ask for "public network IP" and provide a reason.

Most new devices on the internet are wireless phones and most wireless telcos do not give out an ipv4 addresses. They use carrier NAT to share some aspects of ipv4 connectivity but not the important things like being able to use ports.

NAT

Most of those devices only have a private IP address and use NAT to be able to communicate via the Internet.

Only servers that need to be publicly accessed directly like a web server actually need a public IP.

ipv6 is heavily used in the mobile network space.