Backend performance

Backend performance interviews test latency, throughput, and resource efficiency for Node.js APIs and workers. Strong answers name queueing, event-loop blocking, database access shape, pooling limits, and cache consistency—not "add more RAM."

Subtopics: latency, throughput, profiling, database query optimization, connection pooling, backend caching, async bottlenecks.

On interviews: trace a slow endpoint through dependencies; explain p95 vs average; when pooling helps vs when query shape is wrong; how unbounded `Promise.all` hurts the event loop.

Common pitfalls: N+1 queries masked by cache; huge connection pools overloading Postgres; CPU profiles ignored in favor of guesswork; caching without TTL or invalidation.

The trade-off is response-time improvements versus consistency, operability, and database safety.

Checklist:

• Measure end-to-end latency with percentiles.
• Profile CPU, heap, and I/O wait.
• Fix data access before scaling hardware.
• Bound concurrency and pool sizes.