So we could put v4 into a v6 prefix, and v6 hosts would be able to send packets to existing v4 hosts -- this would work just fine. But those v4 hosts could never respond.
If A has a long address and B has a short one: Correct. But IPv6 is lingering 20 years already so there would be no v4-only hosts anymore.
But you could upgrade B and it would work without changing the configuration of A!!!!!
Did you finally get it now? You could upgrade B without touching A and NOT CHANGE the address of either A nor B and it would all work just fine.
With IPv6 you would have to upgrade B, get a completely different address for B and therefore also change all computers that want to access B (either indirectly by DNS or directly by using fixed addresses).
Imagine a network with IPv4 hosts A, B, C, D. If you don't get a new IPv4 address you could add E with a long address and just upgrade those computers that need to access it. It would work all without any changed addresses for A, B, C and D.
Now compare that with the nightmare of getting completely new addresses for A through D, reconfiguring firewalls, DNS entries, scripts, etc. etc. for all combinations. It's not practicable. Any IPv4 workaround, no matter how ugly, is preferrable to a dual-stack scenario.
Changing ANY field in the IP packet format would require a firmware/software update/recompilation of all network stacks running out there, down to the user programs that just want to open a socket to connect to some server.
Or do you think the addrlen argument passed to connect(3) will just grow by magic in all programs, just because they know it's not the ugly IPv6, but the old, trusty IPv4, just extended with an extra byte?
Of course not.
But it would allow to keep using the configuration (DNS configruration for webservers and email-servers, etc.)
Sometimes I think people just refuse to understand the point.
But that's fine by me. Keep celebrating your "IPv6 world day" for the next 30 years.
How did you ever get the time or patience to configure IPv4?
That's a fair question and I give you a fair answer:
By adding one piece at a time. Not by throwing out everything and starting from scratch.
2: If IPv6 were backwards-compatible, we wouldn't. We could go from IPv4 to IPv6 just like going from CDs to DVDs to BluRay. But it isn't and therefore we won't ever replace that structure.
Like I said elsewhere, they are parallel tracks, not assets. If your new Windows 10 computer doesn't run your old Windows XP software, backwards compatibility is a problem. If you need to connect to someone using Skype or FaceTime, whether you do it via IPv4 or IPv6 is irrelevant, since it's transparent to you. However, it is very relevant to network engineers who have to live with problems around NAT, security, inadequate #addresses and so on.
Yeah and so?
Fact is that IPv4 is neccessary to use the Internet in a meaningful way.
Fact is also that IPv6 does not give you any added value. All the added value is only available after that "everybody switches" and we can phase out IPv4, which will never happen.
Nobody said that IPv4 was perfect or even good. But it works and IPv6 does not. (When I can only reach 10% of servers with IPv6 then the Internet does not work, period.)
Uh no. The moment you added even 1 bit to the address, every piece of equipment that has the IPv4 protocol in firmware - a lot of it hard coded - would have needed to be overhauled to support that. Since the IETF saw that there was gonna be an industry-wide overhaul in any case, it did this complete overhaul, tossing in everything learnt in the years of IPv4, so that another IP transition won't be likely in the next 50 years, if ever.
No. For the 10th time: NO.
You could have upgraded piece by piece and without messing up the configuration. Just like when you upgrade your DVD you can still access your CDs. That is the whole point of backwards compatibility.
So, no, absolutely not would there be a "complete overhaul" (which is impossible - and which is why IPv6 will never become mainstream) but instead a piece-by-piece upgrade. By now it would have been completed at least three or four times.
The reason why we all have DVD is because DVD-drives can read CDs.
The reason why we DO NOT have IPv6 is because IPv6 does not understand IPv4 addresses.
It's really that simple. It's not about IPv4 connecting to IPv6 (that would be forward-compatibility, which is impossible in that case) but the other way around.
What I am proposing is that "1.2.3.4" is by definition a valid IPv6 address. But it isn't. IPv6 not only uses a crazy mapping-scheme (which is not automatic) but also a completely different format.
If IPv4 addresses were also valid IPv6 addresses, you could run your server with just ONE configuration for both IPv4 and IPv6.
Of course an existing v4 host would not be able to contact a v6 host.
But the point is that I can replace the v4 host with a v6 host without changing configuration.
That means you can upgrade to v6 with zero cost. No new addresses, no new configuration, no double-stack maintenance, no nothing: Just upgrade your operating system and you're done.
IPv6 is already 20 years (or so) old, by now everybody would be on v6 and v4 would have been a distant memory already.
1: Yes, but once is one time too often.
2: If IPv6 were backwards-compatible, we wouldn't. We could go from IPv4 to IPv6 just like going from CDs to DVDs to BluRay. But it isn't and therefore we won't ever replace that structure.
3: The world will be dual-stack (actually more single IPv4 stack) forever. You will never reach the tipping point where IPv4 servers are so few that they no longer matter.
Does DHCP also create a second set of DNS configurations for my servers?
You pretend as if server administration does not exist.
You kind of want it both ways: You pretend that "IP addresses are running out" (but they don't. There are about 700 addresses per server) and then you say that you can replicate the IPv4-NAT infrastructure (where only one IP address is needed per NAT network) with IPv6.
It will probably hurt you, but I'll say it anyway: If you use DHCP, you also don't need IPv6 because DHCP is (usually) needed only for small networks where 10.0.0.0 offers more than enough address space. So no IPv6 needed there.
I tested a few things with wget and it really illustrates my point:
wget $ADDRESS will work for domain names and IPv4 addresses, but fail for IPv6 addresses
wget [$ADDRESS] will work for IPv6 addresses but fail for everything else.
So for that line of code you will have to create some function that first has to check against IPv6 addresses.
Of course it's trivial to fix a one-line shell script, but these things are buried in huge codebases that need to be adapted. To fix such a problem can take a long time and the fix may cause other problems and new bugs. So why do it when you don't have to?
Basically a IPv4-address interchangeable with a domain name. It is guaranteed not to contain semicolons. It can be used in scripts, URLs, everywhere without any problems.
But IPv6 addresses are special and need a special incompatible syntax. It is not only incompatible to IPv4 addresses, it is also incompatible to the URL-format and countless other formats that use domain names.
Exactly these things are the reason why Google failed to offer IPv6 for google.com (AFAIK they managed to do it, but they had their ipv6.google.com as a workaround for YEARS because IPv6 somehow interfered with IPv4).
The whole infrastructure works with DNS-names and IPv4-addresses which both are guaranteed not to contain semicolons. Introducing semicolons on purpose is the height of stupidity here.
IP6 was in many ways designed to solve problems in IP4- not just address space issues. Thus a break. Because you had to break IP4 to fix it. At this point IP4 is a patched frankenstein protocol with lots of holes, bad implementations, and quasi adherence to RFCs.
Extending IP4 address space would not solve the problems. A new model is needed.
Wrong.
An extended IP4 address space (in the same format) would solve most problems:
- You could upgrade to IPv6 without changing your IPv4 addresses
- You could do it without changing configuration
- You would not have to maintain parallel-configuration
- You continue to use all that software that expects IPv4 addresses, for example "wget $IP_ADDRESS" will fail for IPv6 addresses, because you need square-brackets.
And you could also have your "new model". The "new model" is not the problem. Address-incompatibility is.
Millions/billions of people will not change the address of all their devices. Period. Get it in your head already.
Any scheme that tries to force all people to change their addresses will fail. And it doesn't matter how great it otherwise is. If the addresses are not compatible, it just will not happen, end of story.
In other words: the IPv4 patched frankenstein protocol is superior to the IPv6 protocol, because it allows people to keep their current infrastructure running and does not require new addresses. Any advantage IPv6 has cannot even remotely outweight that flaw. It doesn't even come near it in convenience and practicability.
Seriously, what are you suggesting? That everybody keep their frankenstein IPv4 infrastructure (because they need it) and maintain a parallel "clean" IPv6 infrastructure because it may offer some advantages in the far future? IPv4 is doing fine and according to netcraft the Web stopped growing about 4 years ago. So it will be fine in 4 years and maybe in 40.
To add IPv6 I would have to:
- Check whether all my hardware is compatible to it, down to the Rasberry Pi that I'm using
- Replace all the hardware that is not compatible
- Invent and remember new addresses for all fixed-IP devices that I use on my internal network
- Make sure IPv4 still works
- Test it all
"several days" is pretty optimistic. IPv6 would probably take several weeks for me.
Also "make sure IPv4 still works" is not so easy as it sounds. For example Google failed to do that several years ago and had to create different domain names to make it work. (Although maybe they have fixed that in the last years)
Now if even Google takes several years to just make google.com take IPv6 requests, then you can't call people "incompetent" if they can't do it in several days.
What you IPv6-people just don't understand is that it's much easier to just add a little feature (like port forwarding) to a setup that works and is stable instead of throwing it all away and starting from scratch. Because that is what IPv6 requires.
Imagine a phone company with 6 digit numbers which wants to give users world-accessible phone-numbers. What did the phone companies do? Easy: Just add prefixes to the numbers and everybody is happy. The old numbers stay valid, you can still connect within the old network(s), nobody has to remember new numbers.
But what if phone-numbers would have been expanded the "IPv6-way"?
Then you would have your old number and would receive a completely different new number, which would also be in an incompatible format (maybe letters instead of digits). Then you would have to update all your phone numbers everywhere, to "switch over".
of course such a scheme would fail instantly and that's why IPv6 continues to fail.
The IPv6 adherents just don't get it. If the IPv6-designers were smart enough to just extend the IPv4-address space we would all be running IPv6 already, because it would require no reconfiguration of routers, no reconfiguration of DNS names, no reconfiguration of anything.
But these morons thought that a billion people will just change all their addresses just because they tell them. Well, it doesn't work that way.
- give all your devices new addresses (because these morons didn't expand the address space like any sane person would, they replaced the address space)
- configure all your network infrastructure to manage the new addresses
- maintain two sets of addresses for the forseeable future
IPv6 is broken because it is incompatible to IPv4.
IPv4 at its maximum would be 4 billion addresses - that's it!!! That is just marginally more than the world's population.
Who cares about that?
Currently, the web has stagnated at around 1 billion hostnames (since about a year), but at less than 200,000 active sites (since about 4 years).
All these parked domains and even most active sites share their address with other domains and according to netcraft there are only about 5 to 6 million computers which server all these domains.
Sorry for all IPv6-adherents, but these are facts:
1: The web stopped growing 4 years ago and is unlikely to grow a lot in the future
2: There are currently about 700 IP addresses per server and all others can use NAT, so there is no real shortage of IPv4 addresses.
3: IPv6 is no alternative because it is incompatible (even the address format is incompatible!) and offers zero benefits because you need IPv4 anyway.
Slashdot Mirror
So we could put v4 into a v6 prefix, and v6 hosts would be able to send packets to existing v4 hosts -- this would work just fine. But those v4 hosts could never respond.
If A has a long address and B has a short one: Correct. But IPv6 is lingering 20 years already so there would be no v4-only hosts anymore.
But you could upgrade B and it would work without changing the configuration of A!!!!!
Did you finally get it now? You could upgrade B without touching A and NOT CHANGE the address of either A nor B and it would all work just fine.
With IPv6 you would have to upgrade B, get a completely different address for B and therefore also change all computers that want to access B (either indirectly by DNS or directly by using fixed addresses).
Imagine a network with IPv4 hosts A, B, C, D. If you don't get a new IPv4 address you could add E with a long address and just upgrade those computers that need to access it. It would work all without any changed addresses for A, B, C and D.
Now compare that with the nightmare of getting completely new addresses for A through D, reconfiguring firewalls, DNS entries, scripts, etc. etc. for all combinations. It's not practicable. Any IPv4 workaround, no matter how ugly, is preferrable to a dual-stack scenario.
Changing ANY field in the IP packet format would require a firmware/software update/recompilation of all network stacks running out there, down to the user programs that just want to open a socket to connect to some server.
Or do you think the addrlen argument passed to connect(3) will just grow by magic in all programs, just because they know it's not the ugly IPv6, but the old, trusty IPv4, just extended with an extra byte?
Of course not.
But it would allow to keep using the configuration (DNS configruration for webservers and email-servers, etc.)
Sometimes I think people just refuse to understand the point.
But that's fine by me. Keep celebrating your "IPv6 world day" for the next 30 years.
Yeah but you could not reuse the IPv4 configuration which was my point.
They already did that:
baldur@ballerup1:~$ ping6 ::101.102.103.104
PING ::101.102.103.104(::101.102.103.104) 56 data bytes
So you can set up a computer that has "::101.102.103.104" and no other address as IPv6 address?
I don't think that's possible. This looks just like a wrapper to IPv4 to me.
1: Yes, but once is one time too often.
How did you ever get the time or patience to configure IPv4?
That's a fair question and I give you a fair answer: By adding one piece at a time. Not by throwing out everything and starting from scratch.
2: If IPv6 were backwards-compatible, we wouldn't. We could go from IPv4 to IPv6 just like going from CDs to DVDs to BluRay. But it isn't and therefore we won't ever replace that structure.
Like I said elsewhere, they are parallel tracks, not assets. If your new Windows 10 computer doesn't run your old Windows XP software, backwards compatibility is a problem. If you need to connect to someone using Skype or FaceTime, whether you do it via IPv4 or IPv6 is irrelevant, since it's transparent to you. However, it is very relevant to network engineers who have to live with problems around NAT, security, inadequate #addresses and so on.
Yeah and so? Fact is that IPv4 is neccessary to use the Internet in a meaningful way. Fact is also that IPv6 does not give you any added value. All the added value is only available after that "everybody switches" and we can phase out IPv4, which will never happen. Nobody said that IPv4 was perfect or even good. But it works and IPv6 does not. (When I can only reach 10% of servers with IPv6 then the Internet does not work, period.)
No. For the 10th time: NO.
You could have upgraded piece by piece and without messing up the configuration. Just like when you upgrade your DVD you can still access your CDs. That is the whole point of backwards compatibility.
So, no, absolutely not would there be a "complete overhaul" (which is impossible - and which is why IPv6 will never become mainstream) but instead a piece-by-piece upgrade. By now it would have been completed at least three or four times.
The reason why we all have DVD is because DVD-drives can read CDs.
The reason why we DO NOT have IPv6 is because IPv6 does not understand IPv4 addresses.
It's really that simple. It's not about IPv4 connecting to IPv6 (that would be forward-compatibility, which is impossible in that case) but the other way around.
What I am proposing is that "1.2.3.4" is by definition a valid IPv6 address. But it isn't. IPv6 not only uses a crazy mapping-scheme (which is not automatic) but also a completely different format.
If IPv4 addresses were also valid IPv6 addresses, you could run your server with just ONE configuration for both IPv4 and IPv6.
But the point is that I can replace the v4 host with a v6 host without changing configuration.
That means you can upgrade to v6 with zero cost. No new addresses, no new configuration, no double-stack maintenance, no nothing: Just upgrade your operating system and you're done.
IPv6 is already 20 years (or so) old, by now everybody would be on v6 and v4 would have been a distant memory already.
1: Yes, but once is one time too often.
2: If IPv6 were backwards-compatible, we wouldn't. We could go from IPv4 to IPv6 just like going from CDs to DVDs to BluRay. But it isn't and therefore we won't ever replace that structure. 3: The world will be dual-stack (actually more single IPv4 stack) forever. You will never reach the tipping point where IPv4 servers are so few that they no longer matter.
For the last 20 years it's always: "to be ready" for the great switch that never comes.
But the fact that the IPv6-network is practicably useless means that there will be no non-negligible number of IPv6-only users - ever.
You pretend as if server administration does not exist.
You kind of want it both ways: You pretend that "IP addresses are running out" (but they don't. There are about 700 addresses per server) and then you say that you can replicate the IPv4-NAT infrastructure (where only one IP address is needed per NAT network) with IPv6.
It will probably hurt you, but I'll say it anyway: If you use DHCP, you also don't need IPv6 because DHCP is (usually) needed only for small networks where 10.0.0.0 offers more than enough address space. So no IPv6 needed there.
Exactly. But that reason is very important.
I don't run IPv6 because I have to run IPv4 anyway and I don't want to run two sets of configurations.
Too bad because without that there is no reason to ever switch to IPv6
Take IPv6 as it is, and define that 0.0.0.0.a.b.c.d (or similar) is at the same time a valid IPv4 and IPv6 address.
It's not a technical problem, it's administrative.
Either these devices are used over WLAN (i.e. for home and office use), then you already have a NAT.
Or these devices are connected over GSM, then your telephone company will provide the NAT for you.
So what use-case are you talking about?
But yes, my NAS and some other devices also have fixed addresses at home and I like it that way.
BTW what you are talking about is DHCP and not DNS.
I tested a few things with wget and it really illustrates my point:
wget $ADDRESS will work for domain names and IPv4 addresses, but fail for IPv6 addresses
wget [$ADDRESS] will work for IPv6 addresses but fail for everything else.
So for that line of code you will have to create some function that first has to check against IPv6 addresses.
Of course it's trivial to fix a one-line shell script, but these things are buried in huge codebases that need to be adapted. To fix such a problem can take a long time and the fix may cause other problems and new bugs. So why do it when you don't have to?
Basically a IPv4-address interchangeable with a domain name. It is guaranteed not to contain semicolons. It can be used in scripts, URLs, everywhere without any problems.
But IPv6 addresses are special and need a special incompatible syntax. It is not only incompatible to IPv4 addresses, it is also incompatible to the URL-format and countless other formats that use domain names.
Exactly these things are the reason why Google failed to offer IPv6 for google.com (AFAIK they managed to do it, but they had their ipv6.google.com as a workaround for YEARS because IPv6 somehow interfered with IPv4).
The whole infrastructure works with DNS-names and IPv4-addresses which both are guaranteed not to contain semicolons. Introducing semicolons on purpose is the height of stupidity here.
And you still think that's a great idea?
IP6 was in many ways designed to solve problems in IP4- not just address space issues. Thus a break. Because you had to break IP4 to fix it. At this point IP4 is a patched frankenstein protocol with lots of holes, bad implementations, and quasi adherence to RFCs.
Extending IP4 address space would not solve the problems. A new model is needed.
Wrong.
An extended IP4 address space (in the same format) would solve most problems:
- You could upgrade to IPv6 without changing your IPv4 addresses
- You could do it without changing configuration
- You would not have to maintain parallel-configuration
- You continue to use all that software that expects IPv4 addresses, for example "wget $IP_ADDRESS" will fail for IPv6 addresses, because you need square-brackets.
And you could also have your "new model". The "new model" is not the problem. Address-incompatibility is.
Millions/billions of people will not change the address of all their devices. Period. Get it in your head already.
Any scheme that tries to force all people to change their addresses will fail. And it doesn't matter how great it otherwise is. If the addresses are not compatible, it just will not happen, end of story.
In other words: the IPv4 patched frankenstein protocol is superior to the IPv6 protocol, because it allows people to keep their current infrastructure running and does not require new addresses. Any advantage IPv6 has cannot even remotely outweight that flaw. It doesn't even come near it in convenience and practicability.
Seriously, what are you suggesting? That everybody keep their frankenstein IPv4 infrastructure (because they need it) and maintain a parallel "clean" IPv6 infrastructure because it may offer some advantages in the far future? IPv4 is doing fine and according to netcraft the Web stopped growing about 4 years ago. So it will be fine in 4 years and maybe in 40.
- Check whether all my hardware is compatible to it, down to the Rasberry Pi that I'm using
- Replace all the hardware that is not compatible
- Invent and remember new addresses for all fixed-IP devices that I use on my internal network
- Make sure IPv4 still works
- Test it all
"several days" is pretty optimistic. IPv6 would probably take several weeks for me.
Also "make sure IPv4 still works" is not so easy as it sounds. For example Google failed to do that several years ago and had to create different domain names to make it work. (Although maybe they have fixed that in the last years)
Now if even Google takes several years to just make google.com take IPv6 requests, then you can't call people "incompetent" if they can't do it in several days.
What you IPv6-people just don't understand is that it's much easier to just add a little feature (like port forwarding) to a setup that works and is stable instead of throwing it all away and starting from scratch. Because that is what IPv6 requires.
Then the address would change every time the hardware changes.
To illustrate, let's look at phone numbers.
Imagine a phone company with 6 digit numbers which wants to give users world-accessible phone-numbers. What did the phone companies do? Easy: Just add prefixes to the numbers and everybody is happy. The old numbers stay valid, you can still connect within the old network(s), nobody has to remember new numbers.
But what if phone-numbers would have been expanded the "IPv6-way"?
Then you would have your old number and would receive a completely different new number, which would also be in an incompatible format (maybe letters instead of digits). Then you would have to update all your phone numbers everywhere, to "switch over". of course such a scheme would fail instantly and that's why IPv6 continues to fail.
The IPv6 adherents just don't get it. If the IPv6-designers were smart enough to just extend the IPv4-address space we would all be running IPv6 already, because it would require no reconfiguration of routers, no reconfiguration of DNS names, no reconfiguration of anything.
But these morons thought that a billion people will just change all their addresses just because they tell them. Well, it doesn't work that way.
IPv6 requires you to:
- give all your devices new addresses (because these morons didn't expand the address space like any sane person would, they replaced the address space)
- configure all your network infrastructure to manage the new addresses
- maintain two sets of addresses for the forseeable future
IPv6 is broken because it is incompatible to IPv4.
Who cares about that?
Currently, the web has stagnated at around 1 billion hostnames (since about a year), but at less than 200,000 active sites (since about 4 years).
All these parked domains and even most active sites share their address with other domains and according to netcraft there are only about 5 to 6 million computers which server all these domains.
Sorry for all IPv6-adherents, but these are facts:
1: The web stopped growing 4 years ago and is unlikely to grow a lot in the future
2: There are currently about 700 IP addresses per server and all others can use NAT, so there is no real shortage of IPv4 addresses.
3: IPv6 is no alternative because it is incompatible (even the address format is incompatible!) and offers zero benefits because you need IPv4 anyway.