490
CGNAT version 2
(sh.itjust.works)
Welcome to Programmer Humor!
This is a place where you can post jokes, memes, humor, etc. related to programming!
For sharing awful code theres also Programming Horror.
stupid question, wouldn't it be easier to just have sub addresses?
like my fictional ip address is 123.123.123
and I can set my router to give up to 1000 sub addresses, so one computer can host a Minecraft server at 123.123.123.001 I have another for my some projects, the projects ones each have sub addresses like 123.123.123.002.001 and 123.123 123.002.002...
a company could have countless layers and any amount of addresss they want.
and we're never going to run out of addresses.
Well the IPv4 spec only allows 4 octets, so having 5 or more is impossible. We could fix it by changing the protocol, but at that point it's more worth it to just migrate to IPv6.
fair, but isn't IPv6 just going with the same assumption as IPv4, "so many addresses, no way we will ever use them all"
They didn't read xkcd 865, I guess. Or maybe they did.
there's always a relevant XKCD
It's pretty hard to overstate just how many addresses are in the ipv6 address space vs ipv4.
One of my favorite descriptions comes from Beej's guide to network programming, something I first read probably in the early to mid 2000s. https://beej.us/guide/bgnet/html/#ip-addresses-versions-4-and-6
For every IPv4 address, IPv6 has 18 quintillion IPv4 Internets.
But, sure, it might be possible for us to fsck up allocations, again.
wouldn't surprise me if we end up in a situation where individual programs have their own IP. then individual variables, so different programs in different networks can access them.
that might actually end up consuming all the addresses ...
stupid suggestion. just saying that future technologies might figure up a way to fuck this up again
There's enough V6 addresses for every atom on the planet and enough spare to do it 100x over. We'll be fine.
going to gave each atom in the solar system its own IP address.
checkmate
Yeah, the Universe keep making bigger fools (of us all). But, we should still use IPv6 instead of clawing the tattered remains of IPv4. I just wish my ISP agreed.
without a doubt ipv6 is an improvement. only loss is that it's humanely possible to remember ipv4 addressed, but that ain't necessary.
my only "objection" is that an actual solution should accommodate unlimited growth, rather than what we consider a big enough number.
I think that's a bad objection. It's idealistic in the worst way, it's making "Perfect [...] the enemy of the good". Plus, there are significant practical advantages to a fixed-length addressing scheme, and any fixed-length going to have a maximum. So, under the constraint of fixed-length addressing "big enough" is all we have.
128 bits really is quite hard to fill up, we'll have to worry about a lot of very different things before the run out of addresses. Like speed-of-light latency vs. TCP (and possibly TLS session) timers for interplanetary connections.