New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Careful there, even routing only 2^64 networks (if those will ever exist; I have my doubts) is more than all of today's network to the square. And while I do not think there will ever be 2^64 networks in use if the standard says that's what is possible then routers must be able to handle 2^64 networks.
Also keep in mind that 2^64 times 16 is vastly larger than 2^24 times 4 which means that routing table memory will need to grow immensely. For the sake of orientation: really large fat servers nowadays have something in the range of 2^44 bytes.
So, the IPv6 address space size is the results of an unfortunate hazard?
(a) It does hurt, although you choose to ignore that.
(b) Why not 256 bit addresses then, or, being at it, 1024-bit addresses? Won't hurt either according to your logic.
Yeah right, and the processor manufacturers have worked on larger word size processors pretty much since day 1 just for the fun of it. Following your "logic" we still could use 16-bit CPUs, albeit with more memory and higher clock speed, because hey, computing 64-bit data with a 16-bit CPU is just as fast as with a 64-bit CPU ...
So, me clueless, intel clueless, AMD clueless, IBM clueless ... as is shown by rcy026's vague handwaving at some of their routers.
(a) that may be perceived as [insert negative adjective of choice] but it is not personal.
(b) I may or may not respect your personal preference, but it's not a rule here.
Pure rhetoric. Well known from "democracy is the best system we have" and from "socialism/communism is the best we can do", accompanied by something like "but you are free to create something better".
Btw, my point isn't about you or organization XYZ liking and/or using IPv6 but about the attempt to force-feed insane crap to all of us.
... or maybe due to still much smaller tables.
And we can of course totally ignore ridiculously blown up IP headers.
I've heard of someone who saw aliens and was abducted by them. That certainly trumps physics.
It does not work like the old way, everyone knows every allocation of every AS and the path to it (more or less). IPv6 is hierarchical, this makes it simpler, besides, every AS will have ONE allocation, not hundreds and possibly thousands if we break down the "subnets" further. Of course, multi-continental actors will probably have more allocations, but not anywhere near the current "subnetting".
IPv6 was designed to be routing-efficient and, while I bet many people will ignore those capabilities, considering same internet penetration and usage, if all routing would be IPv6 today, we would need way less routing power than we need only for the IPv4 part.
Yes, in theory, we can have a table which needs to know EVERY /64, but that will never happen, not even divided by 2^32, IMO, so, no, there is no comparison. This is why in practice IPv6 routers work faster and, while as IPv6 rollout continues the load will greatly change, I predict the IPv4 routers will still need more resources than IPv6 ones until IPv4 is largely out.
If you want to be naughty, you can also design the mask in decimal, it doesn't matter, just like IPv6 is also a binary mask.
Personal opinion, the design of 128 bits may have one of the security considerations, such as not easy to be scanned, in addition to easier to configure the address (SLAAC).
You're not making sense. I just explained how you can't make it backwards compatible AND have IPv4 access the new addresses or features. It's pointless and inherits a dozen problems instead of fixing them. You guys don't seem to know what backward compatibility means. IPv6 operates in parallel, it's much more flexible than putting limitations on trying to do stupid backward compatibility.
If you need a real world example, look at how the 2.4GHz band is forever limited due to supporting 802.11b from over 2 decades ago. The 5GHz band skipped 11b and their band isn't gimped to that far back legacy protocol.
In 6GHz, they dropped all that old shit and throughout is now in the gigabits from improved efficiency per bit.
There's several websites where you can read up on half a dozen reasons. Solving the address crisis is easily number one, though.
I'm not sure if you're intentionally being obtuse or just stupid. You're conflating CPU makers wanting to improve performance with what, exactly? You're not making sense. It's like you expect them not to innovate, or something stupid.
I can promise you Intel, AMD and IBM have had IPV6 in their labs from before 2003. Back when expensive routers ran on 32MB flash and 4MB RAM and worked with IPv6 just fine.
The more you keep harping on about this with absolutely no backing from actual manufacturers the more irrational you seem. You don't seem to know how network packets get routed vs processed or else you'd know your whole argument is blown to smithereens by your love of NAT, which is an actual performance bottleneck for IPv4 on low and high resource systems.
That's demonstrably false. You're misinformed on IPv6 and refuse to educate yourself.
Rhetoric? He gave you concrete point. You're the one with false rhetoric.
As one should. Your fixation on them is unhealthy.
Umm, you called physics "fairy tales" last week in defense that the bible was literal and true.
But again, you just dismissed decades of tested evidence and instead brought up aliens. There's just no use with you.
I don't deny there are other reasons, but the main limitation on the internet is bandwidth, if the new protocol included some way to allow more transfer on the same physical lines with some magic, yes, it would have been a serious improvement, nothing that IPv6 came with of any significance could have not been done (and things have been backported) with IPv4.
I never said IPv6 is or was thought to be backwards compatible with IPv4, in fact, my main problem is that it was not. If there would have been an extension of IPv4, like adding more bits and the old IPv4 would have simply gotten some zeroes instead then we would have had the holy grail.
If the designers would have known it would go that slow, I am sure they would have had a different approach.
Yes, because we all know that routers are the only thing processors are used in. Absolutely nothing else, routing is what is driving the CPU market. Things like computers, servers and everything else just uses whatever the routers need.
AMD, Intel and IBM all supports IPv6, so according to you they are all part of the mental asylum so we really should not listen to anything they say.
I think everyone with even a basic knowledge of computing knows that using 128-bit on 64-bit CPU's is not ideal.
But at the same time, anyone with even some basic common sense should be able to by now realize that in the real world this is obviously not a problem since IPv6 routers have been humming along nicely for over a decade with absolutely stellar performance. All the other performance gains in IPv6 is obviously more then capable of making up for the 128-on-64-bit problem.
Pure rhetoric, you are the only one that considers it "insane crap".
If you rephrase it and say "giving us a future-proof solution that already works" it doesn't really sound as bad, does it?
Yes, they do not seem to pose any measurable hit on performance, so ignoring them is just fine. Things like source defragmentation and no IP-level checksum seems to more then well make up for it.
If that person repeatedly and constantly can reproduce the results at any given time, then yes, our understanding of physics would certainly have to be questioned.
Any good scientist knows that if your theory does not comply with reality, you are probably missing something or you are not understanding reality as well as you think. Like your "128-bit-on-64-bit-is-a-big-problem" theory. It works in your head, but it's not a problem in reality.
In China, as long as your router supports it, it will be assigned an ipv6 address of /64, but the routing table is terrible
I'm out of this, as this "discussion" has turned into basically a religious war where provable and proven facts are simply ignored (I'm guilty, too) and increasingly the basic line is purely personal arguments and even attacks.
The main line of some boils down "there is IP4 and IPv6 (and nothing else) and IP4 sooner or later will become unusable, so we HAVE TO USE IPv6". Some even focus mainly on some advantages IPv6 may have for some use cases.
What many seem to fail to see is that that is more of a political than a technical mode of operation (you'll find that basic MOD quite often in many countries and systems).
Personally I have no problem; I'll simply continue to use IP4 only. At home and office the modem (typ. provided by the telecom/ISP) deals with whatever the external interface happens to be and internally I stick to IP4. As for dedis and VPS I'm very confident that I'll be able to get IP4 connected products for many more years.
AFAIC the fact that still nobody could offer a sensible, relevant, and realistic reason/need for a 128-bit address space tells all one needs to know about IPv6.
Proper engineers with a brain would have chosen a 64-bit address space.
To those who felt offended by me I apologize. Have a nice day everyone.
Man, you really have your head far up your ass on this one.
You seriously discard a perfectly working, proven and established protocol simple because you feel that nobody can explain to you why they use 128-bit addresses when you think that 64-bit would be enough?
Your only objection to IPv6 is that the addresses are too long. I know that is extremely simplified, but basically that has so far been the only thing you have been able to point out as a problem and kind of half-assed have a technical reasoning behind. One that seems to have close to zero impact on real world usage, but still, I'll give you credit for getting the theoretical problem valid. In reality, it does not seem to exist at all, but you keep pushing it as a major reason to not accept IPv6.
You could turn it around and say that you still have not offered a single sensible, relevant, and realistic reason as why to not use IPv6.
You don't happen to live anywhere in Sweden, do you? I would love to invite you to some of the NOC's around here and show you when we push 10 and even some 40 Gbit/s links all across the country to the max with IPv6 only, with routers that would have to work a lot harder to do the same with IP4. And we have been doing this for over a decade without any kind of problems at all, so when I say IPv6 works it's not only theoretical.
No.
@jsg even if ipv6 is too long, you will only copy and paste it once into your dns and bang no more problems (that accually exist)
Thanks (I mean it) but my point isn't about what I need to copy/paste/insert.
That's unreasonable demand and expectation. For one thing, that's entirely not the right OSI layer for bandwidth improvements. /head bangs against wall.
whoosh it can't and doesn't work that way.
Armchair network engineering on LET is fairly disappointing considering the users and audience.
You're not qualified to even guess what engineers would do. You fail the brown M&M test nearly every discussion because you refuse to educate yourself.
This is demonstrated when you drone on and on about having so many IP's per person and they only need X amount. I tried to point you in the right direction, but you flat out refuse to learn anything you disagree with.
If you talk to Network Engineers whose jobs are to plan large scale networks, they give numerous reasons for such large addresses. IPv6 is for long term, decades and more.
My town has grown by 10,000 in the last 20 years. The next town over doubled in size and will double again within 20 years.
Single homes get replaced with apartments.
Three story apartments get replaced with 30 story condos.
Farms getting replaced with 10,000 home subdivisions.
For each customer, there's SEVERAL providers offering services on their own networks, not some big shared ISP with one network. So they all need to plan and maintain for future capacity.
There's 7+ billion on this planet with that number going to 11, 12 billion at some point. There's still parts of the world not yet connected to the Internet.
There's many, many more cases. But you're just looking super basic with your lack of knowledge.
Funny. "Mr. f_ck, f_ck, f_ck" doesn't even get the most basic things like "I do not even read his cursing transpirations".
With IPv6 you can assign one address per popcorn kernel in the universe (and on this thread). 🍿
I have assigned a /56 to my rice jar.
Every grain of rice I put in has an address.
Today I ate the xxxx:e4c1:e144:d4c6:4d11:b000/116 bowl of rice.
If I were to do the same on 64-bit addressing jsgnet, I might get a /48 allocation that contains only 65536 addresses.
Then I cannot buy more than 16 days of rice, otherwise my rice jar would explode due to address conflicts.
The guy jumped the shark during the title/benchmark wars and has never been the same.
It can't work now, of course, but it could have worked if IPv4 would have gotten reworked instead of making a new one.
There would have been other issues with that, though, but I think fewer than we have with IPv6 being so divergent.
Yes, it is cumbersome yes, it is more complicated than needed, but this is it and nobody will move to make changes now, whether we like it or not and the rollout will have to be done sooner or later, there is a cost with rolling it out, but the cost of procrastinating in any other case than sitting on a "treasure" of unused IPv4 is higher.
Of course it is, this is why I have mentioned the "magic", the idea is that it did not need to reinvent the wheel for solving a more or less unidimensional problem.
Well its the only reason you've been able to point out so far, and it's not like this is our first IPv6 discussion.
I asked a naughty question~
I have a 1G1C VPS and have a /64, can anyone tell me how to fill up all the addresses quickly?
I don't want to waste every address in the address block, but what can I do?
I have calculated that it is possible to assign /128. to each sand.
Which 2000::/3 approximate size is 42535295865117307932921825928971026432(/128 size), currently about 2000::/9 used, the actual is 585958813449496758547836089653002240 addresses, the Earth's mass is about 5.972e24, I Assuming that each sand is 1mg, the earth is composed of sand, then each sand should be assigned to it /105, it's crazy
However, if the Earth is made of hydrogen atoms(1.7981692e58), it would be difficult to assign an address to each hydrogen atom.
IPv6 feels impractical, is impractical, and is totally oversized. Something very much IP4-like but with 64-bit addresses is what we need - and I've yet to hear a sensible and realistic reason why we allegedly need more IP addresses than molecules in our solar system.
jsg is that you?
What is the purpose of inet6num and route6 for less than /48?
for example:https://apps.db.ripe.net/db-web-ui/query?bflag=false&dflag=false&rflag=true&searchtext=2a05:8980::/112&source=RIPE
https://apps.db.ripe.net/db-web-ui/query?bflag=false&dflag=false&rflag=true&searchtext=2a0c:f00::/112&source=RIPE
No. Its the only reason among others you've been able to spot so far.
Whatever, I'm not interested anymore. Enjoy IPv6 and what you and some perceive as victory ... and have a nice weekend.
Victory? It is a well overdue next step. Call it progress or evolution or whatever, but victory is not a fitting phrase since that implies some kind of competition or battle. There has never been a battle between IP4 and IPv6 in the same way as there has never been a battle between neanderthals and humans, they belong to different eras.
I've tried to explain this to you before, I am not a die hard IPv6 fanatic. I do not think it is an amazing protocol that will solve every problem. If it was up to me, there are things I would have done a lot differently. But it is the only option we have. Clinging on to IP4 for dear life will not solve anything, and I have yet to see any other even close to production ready solution, not to mention one that is already in widespread use and supported by basically everyone but you. We can stay with IP4, dig our heals down and just refuse to use IPv6 because we think the addresses are to long and gain absolutely nothing. We pause progress and just wait for someone else to solve everything in a way that does not hurt our pretty little feelings. We can wait for you to finish your 64-bit IP4+ or whatever you want to call it. Or, we get on the fcking wagon and make something happen, we have had a perfectly working finished solution for decades and people are still whining about how they would have done it and their way would have been so much better. Well, either put your solution into production, get support by major players and make something of it, or accept the solution that has already done all of that and is the new standard, whether you like or not. You literally have deteriorated to the level of "I don't care what the rest of the world runs, I will run IP4 in *my network" level. And I do mean literally, you did say it! I mean, just read that sentence in the voice of a grumpy 4-year old and you'll know what you sound like to the rest of us.
Sorry about the rant, you know I have a lot of respect for you and your experience and technical competence, but man, you are a whiny little bitch sometimes.
He says, right after a post with no fucks at all. Just do your trademark runaway thing when challenged as you have no argument.