mercutio: deutsche telekom is still (one of?) the biggest isps in germany and users might actually blame cf instead of dt for slow connections
mercutio: the problem is also that the other major isps in germany also sucks, e.g. kabel deutschland has weekly (short) outages and several times a year bigger outages and also much congestion in their access layer during peek hours
damn.  yeah, it was more idealism than anything else :)
they'd have to be pretty big to get away with it
they're getting bigger and bigger though.
That cloudflare article isn't being entirely accurate. They are saying that, because they peer 60% of their traffic in Europe, their cost is only the remaining 40%. This only works if peering has zero cost, which it doesn't
mercutio: a friend of mine told me that he couldn't watch youtube with his dt internet connection without buffering but when routes everything trough a vpn to some other country it works fine... (peering between dt and google is a huge problem because google doesn't want to pay dt for transit)
hmm, if google can't move theem, not much hope for cf.
it's good that google are doing that in a way though :)
mercutio: also this is pretty pervert: https://wiki.hetzner.de/index.php/Double_Paid_Traffic/en
well there are fixed costs, and variable costs with bandwidth often plett.
the cost of 10gbe peering is generally pretty cheap though.
wow
that seems a sensible way to go in some ways.
That's what she said!!
i mean to charge customers extra if they want to have more direct connectivity
have you guys seen the nytimes thing about wechat?
http://www.nytimes.com/video/technology/100000004574648/china-internet-wechat.html
it think it's the most sensible thing hetzner can do, but i still think it sucks that dt is holding their clients hostage
yeah
but if google can't change them i don't know who can
bad customer experience?
BryceBot: no
Oh, okay... I'm sorry. 'that seems a sensible way to go in some ways.'
does brycebot simply throw that line out at random, or is there some kind of pattern it triggers on?
sjackso: Bayesian learning, actually.
twss?
That was 10.39% what she said. 'sjackso: Bayesian learning, actually.'
I like it hard
twss?
That was 50% what she said. 'I like it hard'
twss
Okay! twss! 'I like it hard'
twss?
That was 50% what she said. 'I like it hard'
stupid bot :p
But you get the idea.
It's a PHP implementation of this https://github.com/lsblakk/scottbot
(not directly, but the idea of it)
This is the spiffiest speedtest I've seen... ever. /Lots/ of details, and no flash. https://sourceforge.net/speedtest/
brycec: no way to easily share results though
Then you'll just have to trust me when I say I have a 12Tbps connection :D
It's a bummer, but worth it for the detailed results.