Question
Design the caching strategy for a write-heavy counter system: per-post view/like counts for a content platform with 1M writes/sec (increments) but reads that tolerate slight staleness. Write-through to the database on every increment would overwhelm it. Design a write-back / write-behind cache that absorbs increments in memory and flushes aggregates to durable storage, and address the hard parts: what happens to in-memory increments if a cache node crashes before flush, how you avoid losing or double-counting, and how reads see a coherent value.
Clarify scale and constraints first. Propose a clean component breakdown, then go deep on the hard parts — data model, bottlenecks, consistency, failure modes — and name the trade-offs you are making.