This is orthogonal to what causes the pain point. All the pain comes from distributed state and these load levels even if you peak at 800K requests per second you don't need distributed state. So most of this pain is self inflicted.
In most systems I've seen, the caching layer is invalidated more often than necessary, and most of the traffic could've been avoided with a better URL scheme that's more expressive in regards to its content (mutations).
