4 ms·

I'm not super familiar with the development in the low level network space, but I really wonder what the plan for IPv6 is. IPv6 has been the future of IP netwo

by hannob 9y ago

I'm not super familiar with the development in the low level network space, but I really wonder what the plan for IPv6 is. IPv6 has been the future of IP networking for decades. Yet here we are and I still often surf the Internet without v6 connectivity. It suffers from a fundamental problem: As long as IPv6 is not universally available nobody has a real advantage from supporting it. But as long as it provides no advantage many ISPs and web server operators won't bother supporting it. I think this is a design flaw. IPv6 was built without a plausible transition plan. The plan "over time more and more people will switch to v6 and at some point everyone will have it" clearly doesn't work and will never work.

cm2187 9y ago
I guess we will get there. But possibly in 10y. My home broadband in the UK is BT and they just switched on IPv6. Right now it is very unstable. Sometimes I am IPv6 enabled, sometimes not. IPv6 connectivity usually lags behind IPv4 (ie assignment of an IPv6 address is delayed by a few minutes). So it's all hacky and unreliable but it is making some slow progress.
manacit 9y ago
There was a time when running out of IPv4 addresses meant that ISPs and other providers _had_ to switch to IPv6, or they would run out of addresses (or spend too much acquiring them). With NAT and stuff, that risk has seemed to decrease significantly. Google statistics show that nearly 20% of their traffic is over v6, which is honestly not bad. The USA itself is 35%, with no latency hit.
- hannob 9y ago
  > Google statistics show that nearly 20% of their traffic is over v6, which is honestly not bad. The USA itself is 35%, with no latency hit. This is totally bad. We need close to a 100% for v6 to make any sense. 20% IPv6 is pretty pointless.
  - josteink 9y ago
    > This is totally bad. We need close to a 100% for v6 to make any sense. Considering how it was 2% not many years ago, this is a huge improvement. What's going to drive this further is that the cost of supporting IPv4 will increase more and more as IP-space is exhausted and ISPs will have to apply new hacks like CGN. At that point getting people to IPv6 and IPv6-only will be a purely be about economics.
  - Symbiote 9y ago
    For a large ISP, that's 20% less traffic through their very expensive, slow CGN system.

by hannob 9y ago

cm2187 9y ago

I guess we will get there. But possibly in 10y. My home broadband in the UK is BT and they just switched on IPv6. Right now it is very unstable. Sometimes I am IPv6 enabled, sometimes not. IPv6 connectivity usually lags behind IPv4 (ie assignment of an IPv6 address is delayed by a few minutes). So it's all hacky and unreliable but it is making some slow progress.

manacit 9y ago

There was a time when running out of IPv4 addresses meant that ISPs and other providers _had_ to switch to IPv6, or they would run out of addresses (or spend too much acquiring them). With NAT and stuff, that risk has seemed to decrease significantly. Google statistics show that nearly 20% of their traffic is over v6, which is honestly not bad. The USA itself is 35%, with no latency hit.

hannob 9y ago
> Google statistics show that nearly 20% of their traffic is over v6, which is honestly not bad. The USA itself is 35%, with no latency hit. This is totally bad. We need close to a 100% for v6 to make any sense. 20% IPv6 is pretty pointless.
- josteink 9y ago
  > This is totally bad. We need close to a 100% for v6 to make any sense. Considering how it was 2% not many years ago, this is a huge improvement. What's going to drive this further is that the cost of supporting IPv4 will increase more and more as IP-space is exhausted and ISPs will have to apply new hacks like CGN. At that point getting people to IPv6 and IPv6-only will be a purely be about economics.
- Symbiote 9y ago
  For a large ISP, that's 20% less traffic through their very expensive, slow CGN system.