Earlier quoted context omitted.
To do NAT, you need to map (external) port numbers to (internal) IP addresses. This is done using connection tracking: tracking the state of the connection and the appropriate mapping. And connection tracking gets expensive at scale.
Not if you only allow each user 100 connections. That's 1200 bytes of ram per customer paying 100$ a month. And you can charge them an extra $10 per month for 'pro' internet and let them have 1000 connections for 'all the family'.
40% of Google users now connect via IPv6
211–220 of 353 posts
Re: 40% of Google users now connect via IPv6
#212Earlier quoted context omitted.
To do NAT, you need to map (external) port numbers to (internal) IP addresses. This is done using connection tracking: tracking the state of the connection and the appropriate mapping. And connection tracking gets expensive at scale.
Yes, I understand how NAT works. But CPU/memory is going down in cost way faster than bandwidth demand is increasing. Regardless, its way way cheaper than buying IPv4 blocks clearly, otherwise people wouldn't be doing it. Edit: ok, the problem isn't hardware, it's comedy license fees. https://itprice.com/juniper-price-list/cgn.html $470k for a license to do CGNAT at 100gbit/sec. Surely these guys are opening themselv…
Having said that, I’m often equally baffled at just how expensive modern networking hardware is, but as it’s pretty much all of these carrier grade networking solutions being this expensive, I’m assuming it’s somewhat justified.
That doesn’t take away the fact that NAT just adds an expensive layer of complexity on top of it, and I can imagine that in the long term, IPv6 is starting to become much more attractive.
Re: 40% of Google users now connect via IPv6
#213Earlier quoted context omitted.
To do NAT, you need to map (external) port numbers to (internal) IP addresses. This is done using connection tracking: tracking the state of the connection and the appropriate mapping. And connection tracking gets expensive at scale.
Not if you only allow each user 100 connections. That's 1200 bytes of ram per customer paying 100$ a month. And you can charge them an extra $10 per month for 'pro' internet and let them have 1000 connections for 'all the family'.
But that's besides the point. Your home router can easily have millions of connections open (if they didn't skimp on the ram anyway), but if you have CGNAT boxes that do the same for tens of thousands of customers you also have to take into account that they have to move a lot of traffic. This means routing and doing NAT in software won't cut it anymore, but you need dedicated hardware coupled with very fast specialized memory to handle that traffic.
Re: 40% of Google users now connect via IPv6
#214Earlier quoted context omitted.
Yes, I understand how NAT works. But CPU/memory is going down in cost way faster than bandwidth demand is increasing. Regardless, its way way cheaper than buying IPv4 blocks clearly, otherwise people wouldn't be doing it. Edit: ok, the problem isn't hardware, it's comedy license fees. https://itprice.com/juniper-price-list/cgn.html $470k for a license to do CGNAT at 100gbit/sec. Surely these guys are opening themselv…
Software-defined networking is slowly becoming more popular, but it’s always going to be more resource intensive than these enterprise-grade routers that are typically implemented using FPGA / ASICs. Having said that, I’m often equally baffled at just how expensive modern networking hardware is, but as it’s pretty much all of these carrier grade networking solutions being this expensive, I’m assuming it’s somewhat ju…
Re: 40% of Google users now connect via IPv6
#215Earlier quoted context omitted.
With IPv6 each device getting a unique IP is not a bug but a feature -- what will probably happen is that your ISP will lend out a /64 range to you, which your devices will use to assign a unique IP to themselves. This completely removes the need for NAT (also, keep in mind that a NAT is not a firewall or a security feature). BTW, dynamic IP rotation was never a guarantee and is only used because the pool of IPs were…
While I understand the sentiment, NAT does so much to protect most users, it should be considered a security feature.
https://www.anvilsecure.com/blog/dhcp-games-with-smart-route...
https://datatracker.ietf.org/doc/html/rfc2993#page-22
https://threatpost.com/remote-attackers-internal-network-dev...
Re: 40% of Google users now connect via IPv6
#216I sure wish Starlink supported IPv6. They're a brand new ISP planning for millions of customers and decided from the start they couldn't get enough IPv4 addresses for everyone. Fair enough! But we're stuck with Carrier Grade NAT and it is a drag. On Starlink it's impossible to host a server socket directly, which makes any peer to peer networking a PITA. Geocoding IP addresses doesn't work so I have to bend over back…
Re: 40% of Google users now connect via IPv6
#217I sure wish Starlink supported IPv6. They're a brand new ISP planning for millions of customers and decided from the start they couldn't get enough IPv4 addresses for everyone. Fair enough! But we're stuck with Carrier Grade NAT and it is a drag. On Starlink it's impossible to host a server socket directly, which makes any peer to peer networking a PITA. Geocoding IP addresses doesn't work so I have to bend over back…
> A little surprised that a new ISP created in 2021 wouldn't have IPv6 support as one of their launch features. Worse, they had it and turned it off at some point!
Re: 40% of Google users now connect via IPv6
#218Re: 40% of Google users now connect via IPv6
#219Earlier quoted context omitted.
I’m on Vodafone/Kabel Deutschland. IPv4 only, my only option would be to switch to CGNAT to get IPv6, which I’d rather not.
Not sure why thats the case for you. I'm on Vodafone/Kabel Deutschland too and got full ipv4/ipv6 dual stack? Is that one of those regional limitations that Vodafone seems to have?
Re: 40% of Google users now connect via IPv6
#220Earlier quoted context omitted.
To do NAT, you need to map (external) port numbers to (internal) IP addresses. This is done using connection tracking: tracking the state of the connection and the appropriate mapping. And connection tracking gets expensive at scale.
Yes, I understand how NAT works. But CPU/memory is going down in cost way faster than bandwidth demand is increasing. Regardless, its way way cheaper than buying IPv4 blocks clearly, otherwise people wouldn't be doing it. Edit: ok, the problem isn't hardware, it's comedy license fees. https://itprice.com/juniper-price-list/cgn.html $470k for a license to do CGNAT at 100gbit/sec. Surely these guys are opening themselv…
Good. CGNAT needs to die. Addressing is fundamental and customers deserve not just an address, but their own RANGE, especially now that it's feasible.