# Design a Chat Application with a Partitioned Cache Design an end-to-end chat application that supports direct conversations and small group conversations. Cover client connectivity, message-send APIs, durable storage, delivery to online recipients, offline catch-up, ordering, and duplicate handling. Spend additional time on caching: identify which data should be cached, compare process-local memory with a shared cache, choose partition keys, describe cache-miss behavior, and explain how cache invalidation or expiration preserves correctness. ### Clarifying Questions to Ask - Is ordering required globally, per conversation, or only per sender? - Are edits, deletes, read receipts, and attachments in scope? - What acknowledgment means that a send succeeded? - How long can clients be disconnected, and how do they resume? - Which reads are latency-sensitive enough to justify caching? ### What a Strong Answer Covers ```premium-lock What a Strong Answer Covers ``` ### Follow-up Questions - How would you preserve order during a partition migration? - When is an application-process cache preferable to a shared cache? - How would you prevent a cache miss storm after a regional failure? - What changes for very large broadcast channels?