The Trade Desk interview question

Distributed cache design Implement a low latency read APIs supporting millions QPS