Optimizing Qunar's Serverless Platform: Parallel Computation, Thread‑Pool Design, and Cache Governance
This article details how Qunar improved its Serverless FAAS platform by introducing Node.js worker_threads for parallel computation, designing a thread‑pool with shared buffers and postMessage communication, and implementing comprehensive cache pre‑heating and governance, resulting in a 72% reduction in P99 latency and more stable service performance.