Sometimes it takes a shove to get things started. In Netflix’s case, the shove came in the form of a major database corruption, an event that forced Netflix to move from vertically scaled single points of failure, like relational databases in the datacenter, toward highly reliable, horizontally scalable, distributed systems.