It's impressive to see how a company that scales really fast has to be ahead of the curve in how the infrastructure will behave months in advance.
There must be many shaky hands on pulling the trigger on a massive architecture re-write on the assumption that half a year later the current system could go bust.
Interesting for a large and heavy company how fast they've adapted on some of the tech decisions in some cases
It surprises me that PrestoDB only works out for Silicon companies (Facebook, Netflix, DropBox and Uber)