Here's why I like this article. Unlike most commentators on Twitter's problems, Kellan has 1) worked on systems that scale (flickr) 2) is technical and 3) has inside knowledge. He doesn't mention Rails once.
Actually, it's refreshing because despite what most commentators say, most insiders I've talked to don't think Rails has anything to do with the scalability issues. Rails just serves the web pages. If you give it the right data it'll scale just fine. The problem is getting it the right data. You'll notice right now that some features on the web are off and they're all because they're intensive queries in the current architecture.
IM is disabled right now. That doesn't touch Rails in any way.
This isn't to say that Rails has been without problems. I'm sure they've had to make many adjustments and probably dealt with some reliability problems. But I can't think of and haven't heard of a convincing argument for why Rails is the bottleneck to their scaling.
This might be pedantic, but many people also incorrectly say "hockey stick" when they mean exponential growth. Hockey stick growth would be more like a sudden change between two very different rates of linear growth, which would be equally traumatic for your start up in the short term but probably more manageable in the long term.
Hi, I saw this post you made discussing a dissertation that had online dating stats. Could you by chance email me some information about how I could find this dissertation?
