Today in our newest take on “older technology is better”: why NAT rules!
I felt dirty! and broke so much shit when i had to implement NAT on networks in the mid 90’s. Nowdays with ipv6 and getting rid of NAT is much more liberating. The difference is staggering!
- you do not need NAT any longer, firewall is the security, just like on ipv4, just less obscurity.
- you do not need dns views, to workaround NAT any more
- you do not need hairpin NAT to workaround NAT any more
- you do not need to renumber to resize a network. they are always /64, and the answer to how many hosts can it fit is: ALL of them!
- many ALG’s will be unnecessary since there is not NAT.
- vpn’s are easier, since it can be the same address both inside and outside the vpn, the firewall (or host even) enforces the encryption.
- vpn’s are MUCH easier since you will have less rfc1918 collisions due to some other network using the rfc1918 of the vpn’s network
- vpn’s are MUCH MUCH easier since you will have less rfc1918 collisions due to you using the rfc1918 of the vpn partner network, to 1:1 nat a previous vpn network you collided with some months ago… ARGH!!!
- vpn are generally less required, heck i swear 95% of the time the VPN are just to workaround the NAT problem and the data is pointlessly double or triple encrypted.
- you can make more granular firewall rules (eg the spesific host, or network of the source address, instead of the whole enterprise’s public ip) this is real tangible improved security, where any random machine in a network you do not control. do not automatically have openings into your own network.
- firewall objects can if it is suited easily use and depend on FQDN DNS objects when allowing traffic. reducing the need of coordinating firewall object ip address changes between 15 companies.
- firewall rules are easier, more readable, and much more predictable how they will work. All the hairpin nat, public to private nat, private to public nat for a thing that need a different public ip, 1:1 nat for a separate zone, NAT to a vpn or 50 (where 10 of them are 1:1 nat due to collisions, making you require 4 dns views of the same ip space!! ) very quickly gets messy and unreadable. this is probably the largest security benefit. just to reduce the complexity.
- much easier to get people to use dns, since nobody wants to remember ipv6 addresses :D
- nibbles in the ipv6 address can have meanings you assign to them, making the networks and structure both easy to remember and logically structured.
- aggregating routes becomes very easy if you design your network that way.
- firewall policies can become easier if you design your network that way.
- your routing tables is leaner and easier, and of a better consistency. We have 1 large public ipv6 prefix, but 25ish ipv4 prefixes of all kinds of various sizes.
- no need to spend $$ to buy even more ipv4 prefixes.
- no need to have spent hundreds of $$ on a new ipv4 prefix only to be unable to use them for over a year because you need to sanitize the addresses from all the reputation filters. and constantly hound geo ip database providers to update the new country of the prefix. (i am bitter, can you tell…)
- did i mention no need to renumber since you need to grow the /24 to /23 due to to many hosts in a network ?
- did i mention no need to renumber 2 /24’s to /25’s to make space for that larger /23.
- you do not even need any ipv4 addresses any more, use a public NAT64 service, for outgoing. and for incoming just use one of the many free public ipv4 to ipv6 proxies for your services online. for a homelab i really like http://v4-frontend.netiter.com/ (go support them) But most large business l networks use cloudflare, or akamai
- since you do not need your ipv4 address space any more, you can ~~sell them for a profit $$$ ~~ return them to the RIR and give some address space to one of the thousands of companies struggling because they do not have any IPv4 : https://www.ripe.net/manage-ips-and-asns/ipv4/ipv4-waiting-list/
- much lower latency on ipv6, since you do not go across a cloud based ipv4 to ipv6 proxy in order to reach the service ;)
Now the greatest and best effect of ipv6 is none of the above. It is that with ipv6 we have a slim hope of reclaiming some of what made the Internet GREAT in the first place. When we all stood on equal footing. Anyone could host their own service. Now we are all vassals of the large companies that have made the common person into a CGNAT4444 using consumer mindlessly lapping up what the large company providers sees fit to provide us. with no way to even try to be a real and true part of the Internet. Fight the companies that want to make you a eyeball in their statistic, Set up your own IPv6 service on the Internet today !
I felt dirty!
“Senpai, route me like one of your French ISPs”
Sir this is a Wendy’s
JK that’s a lot of good info
i got like a third through it before scrolling to the bottom to see how long it was. omg! should be the canonical example of the opposite of a shitpost ha
This is one of the best comments I’ve ever read on Lemmy. Thanks for writing it. I fully agree with all your points!
Thank you! :) I also notice i compleatly forgot the port exhaustion issue we see with larger networks behind roo few ipv4 NAT addresses…
Con: you are now even more dependent on DNS, increasing the blast radius even more
ifwhen it breaks.But DNS rarely break. The meme about it beeing DNS’s fault is more often then not just a symptom of the complexity of IPv4 NAT problem.
If i should guesstimate i think atleast 95% of the dns issues i have ever seen, are just confusion of what dns views they are in. confusion of inside and outside nat records. And forgetting to configure the inside when doing the outside or vice verca. DNS is very robust and stable when you can get rid of that complexity.
That beeing said, there are people that insist on obscurity beeing security (sigh) and want to keep doing dns views when using IPv6. But even then things are much easier when the result would be the same in either view.
I broke DNS plenty of times in my homelab independent from NAT. In the last few months:
- didn’t turn off DNS server in a wifi router set up as bridged access point
- dnsmasq failing to start because I removed an interface
- dnsmasq failing to start because the kernel/udev didn’t rename an interface on time
- dnsmasq failing to start because hostapd error didn’t set proper interface settings
- forgot to remove static DNS entries in /etc/hosts used for testing
- forgot to remove DNS entries from /etc/resolve.conf after visiting a friend and working on his setup
Yes, most of them is my dumb ass making mistakes, but in the end it’s something that constantly breaks and it helps knowing the IP addresses of my servers and routers.
Aditionally, obscurity is a security helper. The problem is relying only on obscurity. But if I have proper firewall rules in place and strong usernames and passwords I still prefer if you don’t even know the IP addresses of my servers on top of that (in case I break some of the other security layers).
If all that is true, then why do I still hate ipv6 so much.
I’d bet it’s this little bugger " : "
It is for me.
I assume the normal fear of unknown things. It is hard to hate ipv6 once you have equivalent competence in ipv4 and ipv6.
What is localhost now again…
Edit, remember you could use 127.0.0.1, but then it was changed to like 127.0.0.1…something…ff
So guess I was wrong :-) thanks for the info!
::1
For me is because it’s so fucking slow. As soon as I disable ipv6 on every device it has better speeds.
IPv6 is trash.
Tell that to your ISP which has fucked their IPv6 deployment up. In my experience IPv6 is actually faster since it bypasses the IPv4 CGNAT.
On busy days my IPv4 connection can get as slow as 15KB/s, now that’s trash.
Lol that’s ridiculous. There’s nothing about ipv6 that’d make it any slower
There’s one practical thing. Routers have had years to optimize IPv4 routing, which has to be redone for IPv6. Same with networking stacks in general.
In theory, IPv6 should be faster by not having to do bullshit like CGNAT. There’s every reason to think it’ll match that advantage if we just make it happen.
That doesn’t make it “trash”.
But IPv4 addresses are easier to remember!
/s
I could see a point of having home networks stay on IPv4 and NAT with an external v6 address.
That would keep the current security model for home networks where we can assume general tech litteracy is low.
That is not how it works. You can have a home network on ipv6. And it can reach all of ipv4 via nat ( just like ipv4 do today). A net with only ipv4 can not reach any ipv6 without a proxy that terminst the v4 connection and make a new v6 connection. since ipv6 is backwards compatible. But ipv4 is naturally not forwards compatible.
Also it is the default deny of the stateful firewall that always coexist with NAT, since NAT depends on that state, that is the security in a NAT router.
That default deny is not in any way dependant on the NAT part.Interesting, I thought NAT could handle it…
If there is a ipv6 service online. That you want to reach from a v4 only client. You can set up a fixed 1:1 nat on your firewall where you define a fake internal ipv4 address -> destination NAT onto the public ipv6 address of the service. And SRC NAT64 embed your clients internal v4 into the source ipv6 for the return traffic. And provide a internal dns view A record pointing to the fake internal ip record. It would work, but does not scale very well. Since you would have to set this up for every ipv6 ip.
A better solution would be to use a dualstack SOCKS5 proxy with dns forwarding where the client would use the IPv6 of the proxy for the connection. But that does not use NAT tho.
The best solution is to deploy IPv6 ofcourse. ;)
Imagine actually having ipv6 available through your ISP.
…and ever if my ISP actually provided one, getting a static one costs money so there’s no difference in the end.
I guess I am lucky. 3 out of 3 isp’s available from in my region provide IPv6 with a dhcp-pd assigned stable address by default. (Norway)
Yeah, here in Russia the ISPs and IT infrastructure guys seem to be treating IPv6 like it has cooties. I can’t find an article (and it’d be in russian anyway) but as far back as 2022, if you get IPv6 you can expect a variety of issues with it, ranging from “you need to reboot your router every once in a while” to “you technically have v6 but good luck actually browsing v6 internet”.
And of course, why would they give you a stable IP when they can charge for it :T. At least it’s only a third the price of a stable IPv4.
My current ISP technically provides v6 according to their site - but my connection doesn’t have it, and since there’s nothing about it in the years-old contract, I’d need to redo that if I want to complain.
You have my sympathy. I do not know of a sure way to get isp’s to behave. Espesially not if they have regional monopoly
There are usually plenty of choices for ISPs here, actually. But switching between them isn’t likely to give me IPv6 since either they share a magistral or the hardware is just plain old. That, and IPv6 is just not a thing anyone markets.
…and with the current fuckery going on, I doubt many of them have budget for big upgrades. Or maybe even access to hardware to buy.
Fire bad, change scary
Apes together weak
Typing addresses in ipv4 is ingrained into my brain, but zero NATing with ipv6 is magical.
Slightly related to the issue of remembering addresses, I think the main issue is with the fact that local nameservers are pretty much non-existent if you’re not running OpenWrt or OpnSense. Which is shameful because the local nameserver is an amazing quality of life tool.
Also the fact that officially there are no local TLDs except for “.arpa” while browsers won’t resolve one word domains without adding http://
And don’t get me started on TLS certificates in local networks… (although dns01 saves the day)
I don’t get why ‘.local’ isn’t a top level domain for LAN hosts.
I’ve taken to using
.here(or.aqui, “here” in Español, much harder to match outside) as alternatives until something better comes up.Ideally I’d use
.aquí, correctly with the diacritic, but DNS doesn’t seem to support even the basics of Unicode in 2024.Well, there is Punycode, which, if I understand correctly, is a layer before DNS, which translates a Unicode string into a DNS-compatible ASCII string.
I don’t actually recommend using that, though. Every so often, the ugly ASCII string shows up in places, because Punycode translation isn’t implemented there. Certainly increases administration complexity.
Yeah I’ve heard about punycode. Personally, I’m well against it because it puts down non-MURRICAN English domain names as second-class citizens on the internet. If I have a website about Copiapó, a perfectly legal town, there’s no good reason why the domain name should not be
copiapó.clrather thancopiap-xcwhngoingohi4oleleiyho42yt4ptg4ht4.cl, making it look “suspect” and “malware-y”.There were quite some complains back in the time about Firefox choosing not to “flag” internationalized names as potentially dangerous, and pretty much all those complaints that I know of likely came from English speakers who simply can’t understand other countries in the world even can have different alphabets.
I mean, there is some legitimate concerns. For example, in theory, someone could register a domain “αpple.com” and use that to send phishing mails. That “α” is an alpha. The more alphabets and letter variants you allow, the more lookalikes there will be.
But yeah, in practice, domain registrars check that you’re not registering such a lookalike domain and then that’s not really a problem, as far as I’m aware.
.localis already used by mDNSAh, that makes sense. I just knew it was unavailable. Apparently
.lanis fine to use, I think I’ll try that next time.
And don’t get me started on TLS certificates in local networks…
I hate this and the fact that modern platforms seem to require TLS even if you’re serving localhost, so much.
Bro used <> instead of !=
Ha I can remember the ipv6 of cloudflare DNS just fine! It’s uh… something : something : something :: 1111
2606:4700:4700::1111
Hmm, maybe Google is easier:
2001:4860:4860::8888Quad9 is 2620:fe::fe or 2620:fe::9
I don’t understand why they can’t get better addresses than that. Like surely 1::1 would be valid?
Edit: So IANA only control addresses 2001:: and up and there are quite a few IETF reservations within that. I don’t know why they picked such a high number to start at. Everything else seems IETF reserved with a little space allocated for special purposes (link-local, multicast, etc.).
Honestly we should just use 4 bit ip addresses, it’s too hard for me to remember ipv4 addresses anyways. Carrier grade NAT will take care of the rest.
I think it’s worth taking the time to learn IPv6 property. If you have a good understanding of IPv4 it shouldn’t take you more than an afternoon.
Eliminating NAT and just using firewall rules (ie what NAT does behind your back) is incredibly freeing.
I don’t get people complaining about typing out IPs. I like to give all of my clients full FQDNs but you don’t have to. Just using mDNS would be enough to avoid typing a bunch of numbers.
Maybe I have Stockholm Syndrome, but I like NAT. It’s like, due to the flaws of IPv4 we basically accidentally get subnets segmented off, no listening ports, have to explicitly configure port forwarding to be able to listen for connections, which kinda implies you know what you’re doing (ssshh don’t talk about UPnP). Accidental security of a default deny policy even without any firewalls configured. Haha. I’m still getting into this stuff though, please feel free to enlighten me
The problem is we’re projected to run out of unique IPv4 addresses by 2003.
And we are facing the effects of it as we’re speaking. CGNAT and protocols like TURN were not invented without a reason.
Ipv6 is not 6 bytes? 8 segments of 2 bytes for a sum of 16 bytes?
Or am I stupid right now?
Yes, you’ve got it right. <> means ≠. 16 is not equal to 6.
Wait… Do we not like NAT now??
NAT is, and has always been, an ugly hack. Why would anyone like it?
Doesn’t that mean private non-routable subnets like 10.x or 192.x have always been a hack?
No, because there’s use cases for systems that aren’t connected to the internet. Also, public IPs can be dynamic, so you might not want to rely on them internally.
Private addresses don’t necessitate NAT. IPv6 also allows private addresses in the form of
fd00::/8, likefd00:face:b00b:1::1.
You can still NAT IPv6
I haven’t read anything this cursed in a while
Yes, but why would you want to? We have enough addresses for the foreseeable future.
1:1 stateless NAT is useful for static IPs. Since all your addresses are otherwise global, if you need to switch providers or give up your /64, then you’ll need to re-address your static addresses. Instead, you can give your machines static private IPs, and just translate the prefix when going through NAT. It’s a lot less horrible than IPv4 NAT since there’s no connection tracking needed.
This is something I probably should have done setting up my home Kubernetes cluster. My current IPv6 prefix is from Hurricane Electric, and if my ISP ever gives me a real IPv6 prefix, I will have to delete the entire cluster and recreate it with the new prefix.
It should only be needed if your ISP is brain-dead and only gives you a /64 instead of what they should be doing and also giving you a /56 or /48 with prefix delegation (I.e it should be getting both a 64 for the wan interface, and a delegation for routing)
You router should be using that prefix and sticking just a /64 on the lan interface which it advertises appropriately (and you can route the others as you please)
Internal ipv6 should be using site-local ipv6, and if they have internet access they would have both addresses.
Hurricane Electric gives me a /48.
Site-local ipv6 would work here as well, true. But then my containers wouldnt have internet access. Kubernetes containers use Ipam with a single subnet, they can’t use SLAAC.
Point is, you should be able to have them have both. Or stick a reverse proxy in front that can translate. Unless they’re somehow meant to be directly internet reachable the public addresses could be autogenerated
Full disclosure though I don’t know anything about kubernetes.
Yeah, I wonder if there’s any proposals to allow for multiple IPV6 addresses in Kubernetes, it would be a much better solution than NAT.
As far as I know, it’s currently not possible. Every container/Pod receives a single IPv4 and/or IPv6 address on creation from the networking driver.
Even if you give it multiple ip6 cidrs to allocate?
My ISP does this right (provides a /56 for routing), but unfortunately both are dynamic and change periodically. Every time I disconnect and reconnect from the internet, I get a different prefix.
I ended up needing to have ULAs for devices where I need to know the IPv6 address on my network (e.g. my internal DNS servers).
Indeed, that’s correct ula usage, but shouldn’t need nat rewriting. The global prefixes just need to be advertised by RA packets
I use openwrt on my home network which uses dnsmasq for dhcp. It can give a static suffix which just works with the global prefix on the interface and the site local / ula prefix it uses
Note that Android doesn’t support DHCPv6, just in case you have Android devices and ever have to debug IPv6 on them.
So you don’t need to change your network if your isp changes.
You shouldn’t have to?? Maybe you might need to change the mask in your firewall settings if the ipv6 allocation block size changes but that should be it.
Everything else should just work as normal.
IPv6 = second system effect. It’s way too complicated for what was needed and this complexity hinders its adoption. We don’t need 100 ip addresses for every atom on the earth’s surface and we never will.
They should have just added an octet to IPv4 and be done with it.
Every time there’s a “just add an extra octet” argument, I feel some people are completely clueless about how hardware works.
Most hardware comes with 32-bit or 64-bit registers. (Recall that IPv6 came out just a year before the Nintendo 64.) By adding only an extra octet, thus having 40 bits for addressing, you are wasting 24 bits of a 64-bit register. Or wasting 24 bits of a 32-bit register pair. Either way, this is inefficient.
And there’s also the fact that the modern internet is actually reaching the upper limits of a hypothetical 64-bit IPv5: https://lemmy.world/comment/10727792. Do we want to spend yet another two decades just to transition to a newer protocol?
you are wasting 24 bits of a 64-bit register
You’re not “wasting” them if you just don’t need the extra bits, Are you wasting a 32-bit
integerif your program only ever counts up to 1000000?Even so when you do start to need them, you can gradually make the other bits available in the form of more octets. Like you can just define it as a.b.c.d.e = 0.a.b.c.d.e = 0.0.a.b.c.d.e = 0.0.0.a.b.c.d.e
Recall that IPv6 came out just a year before the Nintendo 64
If you’re worried about wasting registers it makes even less sense to switch from a 32-bit addressing space to a 128-bit one in one go.
Anyway, your explanation is a perfect example of “second system effect” at work. You get all caught up in the mistakes of the first system, in casu the lack of addressing bits, and then you go all out to correct those mistakes for your second system, giving it all the bits humanity could ever need before the heat death of the universe, while ignoring the real world implications of your choices. And now you are surprised that nobody wants to use your 128-bit abomination.
You’re not “wasting” them if you just don’t need the extra bits
We are talking about addresses, not counters. An inherently hierarchical one at that (i.e. it goes from top to bottom using up all bits). If you don’t use the bits you are actually wasting them.
you can gradually make the other bits available in the form of more octets
So why didn’t we make other bits available for IPv4 gradually? Yeah, same issue as that: Forwards compatibility. If you meant that this “IPv5” standard should specify compulsory 64-bit support from the very beginning, then why are you arbitrarily restricting the use of some bits in the first place?
If you’re worried about wasting registers it makes even less sense to switch from a 32-bit addressing space to a 128-bit one in one go
All the 128 bits are used in IPv6. ;)
We are talking about addresses, not counters. An inherently hierarchical one at that. If you don’t use the bits you are actually wasting them.
Bullshit.
I have a 64-bit computer, it can address up to 18.4 exabytes, but my computer only has 32GB, so I will never use the vast majority that address space. Am I “wasting” it?
All the 128 bits are used in IPv6. ;)
Yes they are all “used” but you don’t need them. We are not using 2^128 ip addresses in the world. In your own terminology: you are using 4 registers for a 2 register problem. That is much more wasteful in terms of hardware than using 40 bits to represent an ip address and wasting 24 bits.
I have a 64-bit computer, it can address up to 18.4 exabytes, but my computer only has 32GB, so I will never use the vast majority that address space. Am I “wasting” it?
You are using the addressing bits in the form of virtual memory. Right now. Unless you run a unikernel system, then in that case you could be right, but I doubt it.
Anyway, this is apples and oranges. IP addresses are hierarchical by design (so you have subnets of subnets of subnets of …), memory addresses are flat for the most part, minus some x86 shenanigans.
Yes they are all “used” but you don’t need them. We are not using 2^128 ip addresses in the world.
But we do need them! The last 64 bits of your IPv6 addresses are randomized for privacy purposes, it’s either that or your MAC address is used for them. We may not be using those addresses simultaneously but they certainly are used.
Despite that, there still are plenty of empty spaces in IPv6, that’s true. But they will still be used in the future should the opportunity arise. Any “wastage” is artificial, not a built-in deficiency of the protocol. Whereas if we restricted the space to 40 bits, there will be 24 bits wasted forever no matter how.
Hm, didn’t the GP already address (pun unintended!) the 128-bit part?
He/she said the internet is reaching upper limits of 64 bits apparently and gave a value of 61 bits in the linked comment.
64-bit IPv5
64-bit IP would be IPv8, not IPv5.
deleted by creator
it’s not about using all 100 IP addresses for every atom
it’s about having large enough ranges to allocate them in ways that make sense instead of arbitrarily allocating them by availability
