8 ms·

IPv6 is a perfect example of where a legislator can make a positive market impact. I like market principles, but relying on market alone for this just leads to

IPv6 is a perfect example of where a legislator can make a positive market impact. I like market principles, but relying on market alone for this just leads to local optimum (squeezing every last drop from IPv4 with various hacks). Just mandate that every ISP must provide IPv6 for new contracts within a year and then for all existing contract within two - or whatever reasonable timeline - and be done with it.

techsupporter 5y ago
Didn't the US government require that all cloud services provided to and by the government be done over IPv4 and IPv6 some time ago? In opening another tab to answer my own question, I see that they've broadened the scope and that the US government networks should be IPv6-only, with some caveats I'm sure, by 2023: https://fedtechmagazine.com/article/2021/03/how-make-progress-implementing-ipv6-government-perfcon https://fedtechmagazine.com/article/2021/03/how-make-progres...
- ipv6_or_nat 5y ago
  The cynic in me wants to know if all government cloud services support IPv6 as mandated, on paper only or in practice? :) The linked article isn’t as ambitious as to claim IPv6-only government networks by 2023, only that new systems should support IPv6 by 2023. I’d also be very surprised if the government was able to hit any of the IPv6-only targets put forth in the article.
ipv6_or_nat 5y ago
Forcing IPv6 upon ISPs by decree will only make them do the very least possible in order to be able to check a box. To truly drive IPv6 adoption there has to be a real need for IPv6 and/or a revenue driver. Very large ISPs and mobile carriers need IPv6 because they are simply running out of IPv4 addresses, even non-public IPv4 addresses! This still does not mean they can go IPv6-only, they still need IPv4 and translation technologies to reach the Internet. The core problem of IPv6 adoption is that insensitive do not align across the whole Internet ecosystem.
- ClumsyPilot 5y ago
  "Forcing IPv6 upon ISPs by decree will only make them do the very least possible in order to be able to check a box." Which is more than they are doing now. What real-world issues do you forcee. IPv6 either works, or doesn't
  - 2fastpost 5y ago
    > IPv6 either works, or doesn't Oh, you sweet summer child! :) IPv6 “just works” only if you completely control the routing stack end-to-end. Once end user equipment and hosts are introduced into the network all bets are off. In a mobile network, where the carrier controls what user terminals are allowed on the network, it is at least nominally possible to achieve “just works” status. So, in the real world, forcing IPv6 would result in IPv6 being supported only in a particular configuration used in a particular way. Everybody will be told to go pound sand and no effort will be made to keep IPv6 working or prevent breakage.
- tgsovlerkhgsel 5y ago
  "Websites larger than X have to be IPv6-only" would probably do the trick to force adoption. Right now though, we'd be more in need of "websites and -services larger than X have to support IPv6".
  - ipv6_or_nat 5y ago
    This is a nonstarter. Forcing services to go IPv6-only would deprive the majority of Internet users of their services. I can just about imagine the shitstorm and political blowback that would entail. Even just forcing IPv6 support would be a tough row to hoe.
    - tgsovlerkhgsel 5y ago
      If announced with a sufficient amount of time and credibility, it would force ISPs to implement it. It would potentially help them to take it more seriously if there were planned outage windows (i.e. "IPv6-only for 1 day after the law was passed, then IPv6-only for 1 week 6 months after, then permanently IPv6-only after 2 years"). ISPs would probably just take the 1-day outage and ignore customer complaints for a day. By the time the week long outage would come closer without the date being delayed, they'd start taking the issue seriously.
citrin_ru 5y ago
IPv6 is a bad technical standard to begin with (which is on of the reasons behind slow adoption). Forcing it with legislation would only make everyone involved to hate it more.
- throw0101a 5y ago
  > IPv6 is a bad technical standard to begin with Can you expound on why you think this is the case?
  - snuxoll 5y ago
    Not GP, but I think IPv6 is generally fine. The big issue is around mid sized enterprises, where multi homing becomes a need and for so long PI address space was reserved for peering-sizes entities. This makes sense from the perspective of reducing routing table size, but to date there’s no good way to do multihoming with multiple PA blocks (there are ways, but they all suck). The IETF really needs to fix this issue in particular, it’s the last big hurdle in seeing adoption IMO.
    - throw0101a 5y ago
      > but to date there’s no good way to do multihoming with multiple PA blocks (there are ways, but they all suck). Is using ULA and NPTv6 any worse than using NAT44?
      - snuxoll 5y ago
        It’s not - but that doesn’t mean it’s a good solution. Clients should be able to know their globally routable address and ULA with NPTv6 seriously fucks with that, so I’m not a fan.
    - ikiris 5y ago
      PI was never reserved for peering sized entities... just get PI space.
tzs 5y ago
Simpler: put a cap on the number of IPv4 addresses that can be transferred per quarter, and lower that cap each quarter.
- ipv6_or_nat 5y ago
  This just cements the IPv4 incumbents as winners.
  - tzs 5y ago
    I think you might be overlooking network effects (pun intended). Yes, the current incumbents will buy up all the IPv4 addresses they can before trading stops. That will cement them as the permanent winners for IPv4 services. But with IPv4 addresses no longer tradable, hosting companies and ISPs will have no choice but to adopt IPv6 if they want to grow. New hosting companies and ISPs will be IPv6-only. I'd expect at that point most large mobile networks, which are largely IPv6 already with IPv6 compatibility bolted on so customers can reach IPv4-only sites, to go IPv6-only. There will be legacy applications that still need a real IPv4 address on the server side. The current incumbents will end up cemented as the winners of that service.
    - 2fastpost 5y ago
      You assume new entrants or smaller entities would have a fighting chance in this scenario. They would not. The incumbents would become king makers. They would decide which services would be allowed on the IPv4 Internet, which would be the only Internet, and which services would be allowed to prosper. The incumbents would either acquire all successful startups and smaller companies or clone their services. New entrants would wither on the wine on this obscure IPv6-only network that hardly anybody knows about or is able to access. This would also kill off IPv6 completely. The incumbents would have a collective interest in maintaining and propping up their position. IPv6 would be a threat to this, so it would be given the axe. This isn’t some far out fantasy either. This is how mobile phone services worked during the SMS era before the Internet was a thing.
matheusmoreira 5y ago
Don't force IPv6. Make IPv4 illegal. Now they can't half-ass the IPv6 support.
- 2fastpost 5y ago
  Hard pass. Nobody wants more government involvement. Not to mention the unspeakable amount of waste and chaos this would entail.
  - matheusmoreira 5y ago
    If regulation is to be attempted, we gotta do it right. Can't leave loopholes around to make things cheaper for them and worse for everybody else. They should have literally no choice but the right one. Waste and chaos don't matter. Do things right or give up. Letting them keep IPv4 resulted in them dragging their ass, implementing NAT everywhere and killing end-to-end conectivity on the internet, repurposing IPv4 as a reputation system, god knows what else. We gotta get rid of it.
    - ipv6_or_nat 5y ago
      Good regulation is really, really hard. Who do you trust enough to even attempt it? Waste and chaos matter. Or how is your pandemic going?
      - matheusmoreira 5y ago
        If anything, this pandemic proves just how unprepared the world is for such events despite historical knowledge of pandemics. The right thing to do is to devote resources towards preparation for a future pandemic. This is going to cost a lot of money. They won't want to do it.

techsupporter 5y ago

Didn't the US government require that all cloud services provided to and by the government be done over IPv4 and IPv6 some time ago? In opening another tab to answer my own question, I see that they've broadened the scope and that the US government networks should be IPv6-only, with some caveats I'm sure, by 2023: https://fedtechmagazine.com/article/2021/03/how-make-progress-implementing-ipv6-government-perfcon https://fedtechmagazine.com/article/2021/03/how-make-progres...

ipv6_or_nat 5y ago
The cynic in me wants to know if all government cloud services support IPv6 as mandated, on paper only or in practice? :) The linked article isn’t as ambitious as to claim IPv6-only government networks by 2023, only that new systems should support IPv6 by 2023. I’d also be very surprised if the government was able to hit any of the IPv6-only targets put forth in the article.

ipv6_or_nat 5y ago

Forcing IPv6 upon ISPs by decree will only make them do the very least possible in order to be able to check a box. To truly drive IPv6 adoption there has to be a real need for IPv6 and/or a revenue driver. Very large ISPs and mobile carriers need IPv6 because they are simply running out of IPv4 addresses, even non-public IPv4 addresses! This still does not mean they can go IPv6-only, they still need IPv4 and translation technologies to reach the Internet. The core problem of IPv6 adoption is that insensitive do not align across the whole Internet ecosystem.

ClumsyPilot 5y ago
"Forcing IPv6 upon ISPs by decree will only make them do the very least possible in order to be able to check a box." Which is more than they are doing now. What real-world issues do you forcee. IPv6 either works, or doesn't
- 2fastpost 5y ago
  > IPv6 either works, or doesn't Oh, you sweet summer child! :) IPv6 “just works” only if you completely control the routing stack end-to-end. Once end user equipment and hosts are introduced into the network all bets are off. In a mobile network, where the carrier controls what user terminals are allowed on the network, it is at least nominally possible to achieve “just works” status. So, in the real world, forcing IPv6 would result in IPv6 being supported only in a particular configuration used in a particular way. Everybody will be told to go pound sand and no effort will be made to keep IPv6 working or prevent breakage.
tgsovlerkhgsel 5y ago
"Websites larger than X have to be IPv6-only" would probably do the trick to force adoption. Right now though, we'd be more in need of "websites and -services larger than X have to support IPv6".
- ipv6_or_nat 5y ago
  This is a nonstarter. Forcing services to go IPv6-only would deprive the majority of Internet users of their services. I can just about imagine the shitstorm and political blowback that would entail. Even just forcing IPv6 support would be a tough row to hoe.
  - tgsovlerkhgsel 5y ago
    If announced with a sufficient amount of time and credibility, it would force ISPs to implement it. It would potentially help them to take it more seriously if there were planned outage windows (i.e. "IPv6-only for 1 day after the law was passed, then IPv6-only for 1 week 6 months after, then permanently IPv6-only after 2 years"). ISPs would probably just take the 1-day outage and ignore customer complaints for a day. By the time the week long outage would come closer without the date being delayed, they'd start taking the issue seriously.

citrin_ru 5y ago

IPv6 is a bad technical standard to begin with (which is on of the reasons behind slow adoption). Forcing it with legislation would only make everyone involved to hate it more.

throw0101a 5y ago
> IPv6 is a bad technical standard to begin with Can you expound on why you think this is the case?
- snuxoll 5y ago
  Not GP, but I think IPv6 is generally fine. The big issue is around mid sized enterprises, where multi homing becomes a need and for so long PI address space was reserved for peering-sizes entities. This makes sense from the perspective of reducing routing table size, but to date there’s no good way to do multihoming with multiple PA blocks (there are ways, but they all suck). The IETF really needs to fix this issue in particular, it’s the last big hurdle in seeing adoption IMO.
  - throw0101a 5y ago
    > but to date there’s no good way to do multihoming with multiple PA blocks (there are ways, but they all suck). Is using ULA and NPTv6 any worse than using NAT44?
    - snuxoll 5y ago
      It’s not - but that doesn’t mean it’s a good solution. Clients should be able to know their globally routable address and ULA with NPTv6 seriously fucks with that, so I’m not a fan.
  - ikiris 5y ago
    PI was never reserved for peering sized entities... just get PI space.

tzs 5y ago

Simpler: put a cap on the number of IPv4 addresses that can be transferred per quarter, and lower that cap each quarter.

ipv6_or_nat 5y ago
This just cements the IPv4 incumbents as winners.
- tzs 5y ago
  I think you might be overlooking network effects (pun intended). Yes, the current incumbents will buy up all the IPv4 addresses they can before trading stops. That will cement them as the permanent winners for IPv4 services. But with IPv4 addresses no longer tradable, hosting companies and ISPs will have no choice but to adopt IPv6 if they want to grow. New hosting companies and ISPs will be IPv6-only. I'd expect at that point most large mobile networks, which are largely IPv6 already with IPv6 compatibility bolted on so customers can reach IPv4-only sites, to go IPv6-only. There will be legacy applications that still need a real IPv4 address on the server side. The current incumbents will end up cemented as the winners of that service.
  - 2fastpost 5y ago
    You assume new entrants or smaller entities would have a fighting chance in this scenario. They would not. The incumbents would become king makers. They would decide which services would be allowed on the IPv4 Internet, which would be the only Internet, and which services would be allowed to prosper. The incumbents would either acquire all successful startups and smaller companies or clone their services. New entrants would wither on the wine on this obscure IPv6-only network that hardly anybody knows about or is able to access. This would also kill off IPv6 completely. The incumbents would have a collective interest in maintaining and propping up their position. IPv6 would be a threat to this, so it would be given the axe. This isn’t some far out fantasy either. This is how mobile phone services worked during the SMS era before the Internet was a thing.

matheusmoreira 5y ago

Don't force IPv6. Make IPv4 illegal. Now they can't half-ass the IPv6 support.

2fastpost 5y ago
Hard pass. Nobody wants more government involvement. Not to mention the unspeakable amount of waste and chaos this would entail.
- matheusmoreira 5y ago
  If regulation is to be attempted, we gotta do it right. Can't leave loopholes around to make things cheaper for them and worse for everybody else. They should have literally no choice but the right one. Waste and chaos don't matter. Do things right or give up. Letting them keep IPv4 resulted in them dragging their ass, implementing NAT everywhere and killing end-to-end conectivity on the internet, repurposing IPv4 as a reputation system, god knows what else. We gotta get rid of it.
  - ipv6_or_nat 5y ago
    Good regulation is really, really hard. Who do you trust enough to even attempt it? Waste and chaos matter. Or how is your pandemic going?
    - matheusmoreira 5y ago
      If anything, this pandemic proves just how unprepared the world is for such events despite historical knowledge of pandemics. The right thing to do is to devote resources towards preparation for a future pandemic. This is going to cost a lot of money. They won't want to do it.