It's pretty easy to find yourself creating bottlenecked points of failure in systems that expect consistency. There may not be the knobs necessary to scale up p

It's pretty easy to find yourself creating bottlenecked points of failure in systems that expect consistency. There may not be the knobs necessary to scale up past a certain point. A sufficiently bad positive feedback loop can also drive you into the ditch no matter how monstrous your hardware is.