Caching in system design: cache tiers, the limits of Redis, and a 1-million-QPS counter case study
Where caches sit between the client and the database, three jobs (speeding up, buffering, sharing data), the limits of one Redis in numbers, designing like/view counters for a social network at 1 million QPS, unread broadcast notification counts with read pointers, and five small problems solved with Redis data structures.