Eyeballing that chart though it seems to have slowed down in the last half of the year. Also interesting the pattern of slowed down adoption in the beginning of each. I wonder what's up with that.
Maybe people buying new IPv6-capable devices at the end of the year during the holidays and then fewer buying them over the next few months?
It's also interesting to note the difference between weekend and weekday IPv6 traffic (~18% vs ~20%). Businesses are apparently far behind consumers in adopting it.
~10% to ~20% is closer to 100% gain than 10% gain. From 1 in 10 to 2 in 10 people is a doubling of people!
This is why percentage is a bad measure when it can represent two things (percentage or percentage-points).
I like log-odds as a scale to use to represent adoption, going from negative infinity with no adoption through zero (at 50%) to positive infinity at full adoption, but with a typical range of -2 (~1%) to +2 (~99%) when using base e. (Although you can use any base).
If you need to set up even a very small ISP, you will still need to have a relatively sizable v4 ip pool.
Why a radical solution - NAT64 sees much better traction with anybody, but largest ISP's than a "soft landing" solution of 464XLAT favored by big co. is that:
1. Setting up a nat is a 5 minute exercise.
2. It appeals to beancounters because freed IPv4 blocks can be profitably sold.
3. Skype/netmeeting/netflix broke? Not our problem - ISP says. You can see that the proportion of big to small ISPs is much lower outside of USA, and the West in general. And fewer of them bothers with things like direct peering with CDN networks, installing youtube cache appliance or dealing with Neflix openconnect, unless being paid to do so.
Make it mandatory for ISPs to offer leases (at regulated prices) of their last mile to competitors. All of a sudden, the barrier to entry for new wired ISPs drops.
Then, when an ISP fucks up sufficiently, competing ISPs can actually jump in and fill the gap. That last mile is what is building the internet monopolies.
Wasn't that what the US had with DSL? Copper lines had to be made available. Locally, cable was always faster so that's what we went with. DSL was cheaper though. Anyone recall what the experience was like?
Also, they only have to resell ADSL by FCC requirement. However, AT&T voluntarily resells their VDSL Uverse system. I actually went with a reseller in the past because they were still offering no data caps while AT&T was.
In California, what is now AT&T offered retail DSL below the cost to wholesalers, and was very slow to offer 'uverse' (adsl2 and vdsl, for speeds over 6Mbps) to wholesalers. For ADSL1, it was a compelling offering, the data portion of the circuit was delivered to the wholesaler, so they could deliver things like straight Ethernet over ATM, instead of PPP over Ethernet, and had different IP routing than At&t, for uverse, it was basically a normal att account billed to the wholesaler. Getting problems fixed could be problematic because of dealing with one company that had to deal with another; otoh they were experts at dealing with att, because they did it so often.
Also: the FCC decided line sharing didn't apply to cable internet pretty early in the lifetime of cable internet. Then when DSL companies complained (through courts) that that wasn't fair, the FCC said we have enough competition between cable, DSL, and power line (yeah) so DSL companies don't need to line share either; only voice needs to be line shared.
16 comments
[ 1.6 ms ] story [ 45.8 ms ] threadIt's also interesting to note the difference between weekend and weekday IPv6 traffic (~18% vs ~20%). Businesses are apparently far behind consumers in adopting it.
This is why percentage is a bad measure when it can represent two things (percentage or percentage-points).
I like log-odds as a scale to use to represent adoption, going from negative infinity with no adoption through zero (at 50%) to positive infinity at full adoption, but with a typical range of -2 (~1%) to +2 (~99%) when using base e. (Although you can use any base).
By that scale it went from -0.95 to -0.65.
Why a radical solution - NAT64 sees much better traction with anybody, but largest ISP's than a "soft landing" solution of 464XLAT favored by big co. is that:
1. Setting up a nat is a 5 minute exercise.
2. It appeals to beancounters because freed IPv4 blocks can be profitably sold.
3. Skype/netmeeting/netflix broke? Not our problem - ISP says. You can see that the proportion of big to small ISPs is much lower outside of USA, and the West in general. And fewer of them bothers with things like direct peering with CDN networks, installing youtube cache appliance or dealing with Neflix openconnect, unless being paid to do so.
Then, when an ISP fucks up sufficiently, competing ISPs can actually jump in and fill the gap. That last mile is what is building the internet monopolies.
Also: the FCC decided line sharing didn't apply to cable internet pretty early in the lifetime of cable internet. Then when DSL companies complained (through courts) that that wasn't fair, the FCC said we have enough competition between cable, DSL, and power line (yeah) so DSL companies don't need to line share either; only voice needs to be line shared.
The moment someone else offers equivalent speed 50Mbps symmetric Unlimited for $13/Month, I will be out.