IPv6 is fucking stupid. If they wanted it to be accepted they should have just made it like IPv4 but with more numbers and periods at the end.
Also what the fuck happened to IPv5?
Ah yes, why couldn't we have IP addresses like 109.231.12.203.111.224.49.34.154.48.167.253.156.5.9.250
>>109104157The real problem with IPv6 is that it isnt even actually future proof.Why would they only go with 128bits and not 256?
>>109104203That’s exactly what IPv6 addresses look like to me.
>>109104203ngldesu first two numbers look exactly like my home ipv4 address
>>109104242i know
>>109104160This is what I want to know
>>109104212128bits is too much.64 would have been fine for many centuries to come and our great great grandchildren could always invent IPv7.
>>109104242Be careful if we ever find out the other two digits then you're plain fucked.>>109104160>>109104252Existed briefly but it was for something like audio or some shit. Was never supposed to be for the same thing we used IPv4 for. The 6 in IPv6 also stands for how many people thought it was a good idea.
>>109104157The double colon syntax is bad and I never figured out how that expands.Like 2001:0db8::7334
>>109104157*ctrl+c**ctrl+p*heh nothing personal 1337 h4x0rZ
>>109104267nah 256 bit it is. IP addresses should last billions years.
>>109104288Just zeroes in the gap
>>1091042882001:0db8:0000:0000:0000:0000:0000:7334It's just zeroes.A full IPv6 address has 8 groups, 8 - 3 = 5, so you fill in 5 groups of zeros.Of course rarely will you ever need to type the expanded address.
Im just ready for IPv8, these level 4 and 6 spells arent doing anything.
It's stupid that I have to pay to have a personal ipv4 address when I'm already paying for the internet connection
>>109104288Anon it's not that hard. Double colons mean all zeros between them. You take the greatest amount of consecutive zeros and do that to shorten it.>2001:0db8:0000:0000:0000:0000:0000:7334If it were instead>2001:0db8:0000:0000:0000:0000:0000:0003it'd be 2001:0db8::3and if it were>2001:0db8:0000:0000:0000:0000:9000:0011it'd be 2001:0db8::9000:0011etc.
>>109104317They should have kept all the : and just removed the 0s
>>109104332They should have made your mum your dad.
>>109104332vve need all the zeros, because vve are going to have that many devices on our netvvorks one day
>>109104348Why?
>>109104157ipv6 in decimal notation would be 16 three digit numbers, compared to the 4 of ipv4, would you really prefer all that typing?
>>109104369just let AI do the typing bro
>>109104293>*ctrl+p*it's v, for paste! vaste
>>109104203looks like SNMP
>>109104272>The 6 in IPv6 also stands for how many people thought it was a good idea.LOL
>>109104369>would be 16 three digit numbersNot true because 90 out of 256 are two digits and 10 our of 256 are only 1 digit.You'd need 41 decimal digits on average, and lower if you're a bit smart with assigning commonly typed IP addresses lower numbers.Compared to 32 harder to type (because you need both numbers and letters) hexidecimal digits for IPv6.I will argue the decimal system is faster on average.
>>109104157Why didn't they make them variable length?Assign short addresses to things people need to configure often like office/home internet and longer addresses for shit like mobile phones and IoT.Yeah yeah hardware, but they could easily map human readable variable length addresses to fixes length binary for internal use.Start every human readable address with a two letter country code (US, UK, etc) so small countries can use short addresses and make it super easy to range ban countries with loads of botnets like China.
>>109104212Because 265 bits would add bloat for no good reason when we can't even physically use all the IPv6 address space any time soon, it's a waste of ram, storage, and processing.I know the irony of saying it, but again, we literally CANNOT run out of IPv6 addresses on earth.Right now only 2000::/3 is used on the internet, the recommended prefix size is between /48 to /56 per subscriber (most ISPs give out a /56 and the shit ones give you a /64 prefix).If we split 2000::/3 entirely into /48 networks that would make thirty-five trillion, one hundred eighty-four billion, three hundred seventy-two million, eighty-eight thousand, eight hundred and thirty-two networks, or 2^45, each of those networks has over a septillion individual addresses.That is just the global unicast range (2000::/3) we still have 4000::/3, 6000::/3, 8000::/3, a000::/3, c000::/3, and e000::/3 all of which are reserved for future use, for reference each /3 has 42 undecillion individual addresses, and the entire IPv6 address space has room for 340 undecillion addresses.
>>109104421oi mate thanks for reigniting buried painful memories of implementing nagios plugins in Perl.
>>109104317So you need to remember how many colons there are in every address and count the printed one to supstract them from the total amount to know how many zeroes you have to insert. Who the fuck
>>109104293>ctrl+pZoomers and Alphas these days don't even know keyboard shortcuts cuz they only ever use Touchscreen mobile devices.Sad!
>>109104580Anon, the length of the address never changes. The number of colons will always be the same for an expanded address.
>>109104521All of that math is irrelevant because you're just picturing current state of things. Obviously 128 is enough right now, but it isn't actually future proof. When we need to discard bits to switch protocols or upgrade, then how would we do so? They should've just bitten the bullet and gone to 256.The hardware of the time already couldnt process 128, so why not set a high bar anyways.
>>109104160It got taken out back to tend to them rabbits.
>>109104157>
>>109104790>YEARS OF WORK yet NO REAL-WORLD USE FOUND for going bigger than 10.0.0.0/8
>>109104580>So you need to remember how many colons there are in every addressCorrect. Every address has seven colons. I wish you luck in your studies to remember this.Nobody is forcing you to condense the address if this is too hard for you. But it really is not difficult. You can fill from both ends of the address first and then just put zeros in the middle section when you run out of other values for the start and end.>>109104790>We had a tool for that: it was called NATNAT is gay and becomes extra gay when you actually need to segment larger networks with more firewalls/routers/gateways. Reducing overhead should be one of the first optimizations you make when administrating a network. If you're using Class C addresses (192.168.0.0/16) exclusively then this does not apply to you and NAT is fine. If you use anything more complex then a /24 subnet then you'll start to understand why link-local and global IPv6 is nice to have.>inb4 do you actually NEED it?Nobody is doing static configs of IPv6 for every hextet. Nobody sane is deploying IPv6-only networks for end-users. IPv6 is very useful for networking, but not in a way anyone except a network technician could appreciate. You know, the guys helping build and maintain the backbone infrastructure?
>>109104790Governments wants to kill NAT and make every device uniquely registered to an individual.
>>109104928>If you use anything more complex then a /24 subnet then you'll start to understand why link-local and global IPv6 is nice to have.True. The problem is most consumer ISPs do not properly handle IPv6 so no one uses it because no one wants to have to deal with public private prefix translation bullshit at home.
>>109104946It's funny because if you asked me who was most at fault for causing stagnation within the networking industry it would be ISPs. They just want to sit on the same configurations and hardware for eternity while they suck up all your monthly bills. They have zero desire to improve services outside of maybe installing more IPXs for developing areas, and even then go fuck yourself if your regional cache for google.com is a 90ms route because failover is expensive and we don't give a fuck.I had a local ISP swtich to native IPv6 routing through their ENTIRE backbone infrastructure and I switched over to them. Differences were minimal at first but as they rolled it out more thoroughly I noticed a substantial improvement over medium to long-distance connections.
>>109105018I frankly love to shit on the (at times) horrifically shit parts of the IPv6 spec but ISPs are honestly more to blame than the spec itself even if the greed and incompetence of ISPs should have been a consideration from the start.But seriously once you see a consumer contract that implements regular (I've seen shit as bad as 7 days before) mandatory reconnects on a AON FTTH line with a static IPv4 address but a dynamic pool of IPv6 prefixes and the one delegated to you can change with the reconnects you think to yourself maybe the Cuban missile crisis in the 60s did escalate and this is just hell.
>>109104267At least we actual *have* 64-bit processors.128 is retarded until we have things like a ZAX 128-bit accumulator like we have RAX fir 64 bit.
>>109105088The problem seems to be a rather even split between ISP cuckoldry and the spec itself being annoying. Implemented correctly it could serve a great long-term solution to the expanding regional and continental network, but who the fuck cares if all anyone uses is full of the kind of bullshit you're describing. That's why IPv4 prevails methinks. Not because it's inheritly better than IPv6, but because if we want genuine improvement we need genuine adoption. Feels sort of analgous to EVs needing more charging stations to become more viable. Nobody wants to pay for the infrastructure because nobody is buying EVs because the infrastructure isn't there. That's slowly changing, so I wonder if IPv6 implementation could follow a gradual incline.
>>109104506This. It really should have worked like phone numbers. And they didn't really need that many fucking addresses.>0000::, local loopback>ffff::, anycast>0001:0000::USA hosts>0001:0001::Canada hosts...>0044:0000::UK hosts (+44 is the UK phone prefix)and so on. Instead of memorizing 128 bits of shit it should have been"This server is in America, in AWS, in us-east-1, therefore it's going to be 0001:0000:aaaa:0001" where aaaa is arbitrarily amazon and 0001 is us-east-1. Azure could be 0001 and IBM could be :3270: or something and everything would be predictable and geolocation wouldn't be a thing anymore.
>>109105206At least with EVs we're slowly getting there due to hybrids. There is no transitionary tech between IPv4 and IPv6 in the way that hybrids can lead to infrastructure slowly getting better until full EVs become viable. At least that's what has happened in cities and bigger towns around me. Problem with IPv6 is it often requires you to massively overhaul your network or at the very least complicate your setup just because ISPs fucking LOVE non static prefixes and you have to deal with that somehow.And that's assuming you get at least a /60 assigned to you because again some ISPs have a deep seated need to fuck over their consumer tier customers and assign them really cursed subnet configs. I've seen everything from higher than /64's assigned (yes seriously shared IPv6 assignments) and other providers assigning a fucking /40 per customer.I wish IPv6 wasn't such a cluster fuck in reality. The spec is to blame sure and I love to do that but fuck me do some ISPs go out of their way to implement IPv6 in the worst possible ways.
>>109105278>I've seen everything from higher than /64's assigned (yes seriously shared IPv6 assignments) and other providers assigning a fucking /40 per customer.It should have just worked like IPv4 where one connection is one address and it's up to your bitch-ass to NAT or not NAT.
>>109104212the "problem" only exists in your head
>>109104157IPv6 would have been accepted if it had been introduced before CIDR and NAT. As it is, IPv6 is a solution in search of a problem. Nobody wants IoT or more Indians online.
>>109104212Yea also there is no future. It is over and it is all going to collapse. No one even uses I pee vagina sex.
>>109104935Well, they can eat hot shit.
>>109105288That was the intent of the spec but not what the hard requirements said and here we are..
>>109104157It is. It's just the .'s became :'s and added an extra number to the octet.
>>109105288Except NAT was just supposed to be a stop-gap until they impelemented a solution. The IETF didn't want to rely on a band-aid because as I mentioned before, overhead. Dual-map sockets are cancerous. Hybrid systems are inefficent. Sure, I kind of wish we got classes instead of arbitrary addresses so you could use something like ABCD::0/64 and understand it's a private address, but then the whole fucking point was so we didn't have classes again. You want a direct connection, and properly implemented IPv6 WILL give you the most direct, efficient connection IF the fucking ISPs pulled their fingers out of the asses.CG-NAT is even worse for consumers unless your network knowledge is "that's the box that gives me internet". Make a cloudlfare tunnel all you want, but static public IP is superior.
>>109104288>The double colon syntax is bad and I never figured out how that expands.It's just 0's.2001:0db8:0000:0000:0000:0000:0000:7334 is that address.
>>109104580>how many colons there are in every address 7 (for 8 octets)>count the printed one to supstract them from the total amount to know how many zeroes you have to insertSee above.>Who the fuckNon-retards. So not (YOU).It's not THAT hard, Anon. Once you've seen the "::" (which can only be used ONCE in an address and only used for the highest amount of 00'ed octets) shortcut used on an expanded address, it's not THAT hard to grok it (@GORK IS THAT TRUE?!).I'm not even a network admin and even I understand it, Anon.>>109104790AdachiTRUE.gif
>>109105407>the whole fucking point was so we didn't have classes againBut classes are comfy so we shouldn't have done CIDR or gotten away from them. It's perfectly reasonable to say>Megacorp #1 gets a Class A>Megacorp #2 gets a Class A>smaller corp gets a Class B>Facebook gets 5 Class As>Google gets 20 Class As>Amazon gets 50 Class As>the United States Army gets 100 Class As>the entire continent of Africa has to split up and share 5 Class BsIPv6 solved the wrong problem by being protocol incompatible. We just needed more than 256 Class As to give out.
I still don't understand why they couldn't simply append another byte to v4 and maybe free up all the unused and "reserved" addresses. That alone would be enough for a trillion unique addresses, before NAT.
>>109105490>just append another byte to a 32-bit integer
>>109105390>could have added 4 dots>instead they added 11No wonder it's a total flop.
>>109105506>you can't take a 64 bit integer and only use 40 bitsYou literally could have the classic ipv4 format (a.b.c.d) but expandable and easier to use (a.b for local networks, a.b.c.d for large networks, a.b.c.d.e for public/global networks). It even would keep the "notation" backwards compatible with existing networks.
>>109105506>set the high bits on an IPv4 frame address to indicate "this is a long address, not an int">IP driver looks in a new field at the end of the frame for the address, which is, lets say a 64 bit int.>if you're using a non long-address version of Winsock, the address is valid but unroutable (maybe use 127.0.0.2, for example)This would have worked fine and all the IP code would have been extended incrementally. Best case, a dumb ethernet physical layer switch wouldn't even care about the difference.
>>109105532You would have to replace all the infrastructure hard-coded with the length of an IPv4 header. At that point you might as well just use IPv6.
>>109105552It's more about ease-of-use. I can remember four or five sets of numbers whenever I need to manually type an IP. Nobody can remember the clusterfuck in an ipv6 address.And they are replacing the infrastructure anyway. Or trying to. We're still using v4 with dual stack hacks because nobody wants to deal with v6's shit.
>>109104157I just ban all IPv6 on my websitefuck u
>>109106051Why assign an IPv6 address in the first place then retard?
>>109104421cursed post
>>109104770slap NAT onto ipv6simple as.
>>109104935>source: my ass
>>109105580>what is dns>what is mdns>what is ipamNetwork sysadmin larpers are all retards trying to recall numbers rather than using tooling they have at hand.
>>109106104Exactly my point. IPv4 wasn't enough, so it needed NAT.IPv6 still isn't enough, so it still (eventually) needs NAT.
why do we even have all these IPs?4 billion public IPs and 65 thousand ports, that's 200 trillion connections at the same timehow the fuck is that not enoughmaybe we should cut back on communication through IP and use specific protocols for specific systems
test
>>109104157Luckily I have invented a new protocol...An IPv8 address is a 64-bit value:r.r.r.r.n.n.n.nr.r.r.r -- 32-bit ASN Routing Prefixn.n.n.n -- 32-bit Host Address (identical semantics to IPv4)I'm Indian btw, in case it matters.
>>109106543>how the fuck is that not enoughThere are a lot of devices out there. No, there's more than that.What's more, we need some structure in addresses to make them routable in hardware, so we can't just use any old random number.And some parts are just misallocated but can't be changed without the cooperation of an incumbent who doesn't want to help.tl;dr: we've got a situation where we ought to build something bigger and manage it better.
>>109104369implicit leading zeros, there, now you type the exact same amount for the reserved lan ones.
>>109104770Because you have to stop at some point, 128 was basically a compromise between those who wanted variable length addresses and those who didn't, they chose it because it's practically enough, and while obviously not infinite so long as we don't do stupid shit like trying to encode meaning into IPv6 allocations it SHOULD be future-proof enough for our foreseeable future, which means NOT trying to split the entire address space into continent + country + state + city/town allocations, I'm sure 256 could handle that, but it would still ultimately lead to fragmentation as borders change.256 bit is obviously more future-proof, but that also means IPv6 base headers would need to be 72 bytes, that is an additional 32 bytes added to every single packet which on a global scale is a lot of wasted bandwidth right now (if you pretend everyone is using IPv6), with a VPN you are now using 144 bytes are now just headers, and I think it would be ever harder to convince people it's worth it when there are people who think 128 bits it already too much.I also think that even without NAT 128 bit is enough for the Earth, Moon, Mars, and maybe a couple space colonies provided we aren't stupid with the address allocations, at a certain distance you drop IP anyway for something like IPFS (but good) because of the roundtrip time, also the politics of an earth organization controlling the address space.I'm basically repeating myself, but 265 gives an insane number of addresses, it's also a lot more overhead for an internet protocol that will likely be replaced when we need extra-terrestrial networking (if they are still using TCP/IP but encapsulated in some bullshit to send data to deep space colonies I will literally rise from my grave and haunt them).I feel like no matter the size though we will still probably fuck up the allocations somehow, even beyond 256 bits.
>>109104157Who cares? Why have you developed an opinion on this? Does it bother you to see a 5-digit home address because it could be 3-digit instead?
lol @ the fags crying about a couple lettersin my time ip addresses were 1024bit using ascii85 notation try remembering:9x$R+mQ2v9!A_B3k=Zp7W~vL4qY@mNx9/4fK_92pX;vL$mQ!zRt7W~vL4qY@mNx2a/kQ9A_B3k=Zp7W~vL4qY@mNx9x$R+mQ2v/_B3k=Zp7W~vL4qY@mNx9x$R+mQ2v9!A/zRt7W~vL4qY@mNx2a4fK_92pX;vL$mQ!each empty block could be compressed using *, and consecutive blocks can be replaced with //, also *///832 is reserved for non-broadcast proximity networking, so if i wanted to gene share with your sister locally i would use *//zRt7W~vL4qY@mNx2a4fK_92pX;vL$mQ!we still had to wrap our http addresses in square brackets though
if i had a greenfield network to architect i would use only ipv6
