Early on I remember the heroics that Netflix went through running on EC2, building numbers of complex tools for varied performance, reliability issues, etc. At the time many much larger sites just quietly worked 24/7, minus the heroics, and minus the effort, often by using purpose-suited dedicated servers. Later, once Amazon rolled out SSDs, Netflix triumphantly announced how much of an improvement it was to their product, again demonstrating that they were creating a problem (huge numbers of horrible I/O machines) that they then solved with gusto, albeit unnecessarily.
I'm not trying to be overly down on Netflix, but it is a company that seemed to make tech blog entries a product of the company years ago, and the result is that people have bought into this notion that they're doing some hugely complex task. They aren't.