Node.js at scale, rebalanced: how we cut cost by 38%
By optimizing our biggest Node.js service, we cut our compute costs by 38%: we ran 30% fewer pods, reduced memory usage by 20% per remaining pod, and took up to 10% off latency at p75, p99, and p99. No new hardware, and no code changes in the service itself. The gains came from changes to our Node.js platform and a fresh look at configuration choices we had left untouched for years.