Async bottlenecks

Async bottlenecks in Node.js come from misuse of non-blocking APIs: promise waterfalls, serial `for` loops with await, unbounded concurrency, blocking the event loop with sync crypto/JSON, or slow consumers on streams/queues.

					// Bad: serial awaits
for (const id of ids) await fetchOne(id);

// Better: bounded concurrency
await mapPool(ids, 10, (id) => fetchOne(id));
				

| Smell | Fix | |-------|-----| | Promise waterfall | Parallelize independent steps | | Unbounded `Promise.all` | Semaphore / pool limit | | Sync `fs.readFileSync` | Async APIs or worker threads | | Missing backpressure | Pause streams when buffer full |

Use `performance.eventLoopUtilization` and APM to spot event-loop stall. CPU-heavy work belongs in worker threads or separate services.

On interviews: concurrency limits; difference between parallel and serial async; why `setImmediate` does not fix CPU blocking; queue consumer lag.

Common pitfalls: `await` inside tight loops; mixing sync and async file APIs; ignoring downstream 429/503 as backpressure signals.

The trade-off is simpler sequential code versus throughput and fairness under load.

Checklist:

• Bound parallel async work.
• Keep event loop free of sync CPU spikes.
• Propagate timeouts and cancellation.
• Measure consumer lag on queues.