Concurrent HashMap

Phase 1: Requirements & Use Cases

Interviewer: "Design a thread-safe hash map."

Candidate: "To confirm scope: we're building a concurrent hash map that supports safe concurrent reads and writes from multiple threads — think Java's ConcurrentHashMap at the class level. We're NOT building a distributed key-value store. This is a single-process, in-memory map with concurrency guarantees."

Interviewer: "What use cases are you targeting?"

Candidate: "The core use cases are:

1. get(key) — read a value by key, ideally lock-free
2. put(key, value) — insert or update a key-value pair
3. remove(key) — delete a key
4. containsKey(key) — check for existence
5. resize — grow the internal table when load factor is exceeded
6. iterate — weakly consistent traversal without ConcurrentModificationException"

Interviewer: "What's your target concurrency model?"

Candidate: "High read concurrency with lock-free gets, and fine-grained locking for writes to minimize contention. I'll follow the Java 8 ConcurrentHashMap approach: CAS for inserting into empty buckets, synchronized on the bucket's head node for collision writes. This avoids a global lock entirely."

WHY: A global lock serializes all operations — it's correct but kills throughput under concurrent writes. Per-bucket locking allows N threads to write to N different buckets simultaneously.