My node.js process is on fire

on YouTube ↗25m

" summary

At 10am on Black Friday a Node.js app overload forces autoscaling to quadruple servers and costs spike; investigation workflow is measurable: pick a percentile goal (he uses 91st percentile), reproduce on staging, run an HTTP load with autocannon, collect traces with clinic doctor and clinic flame, inspect flame graphs, and iterate. Examples include finding missing DB indexes showing up as gigabyte heaps and using fast-json-stringify and Pino to cut CPU and latency, with demos and working tools.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 2:59How to fix?
3. 7:56Fixing performance issues
4. 8:59Latency and throughput are connected
5. 11:32AutoCannon

" related talks

• 13mMy Node.js process is on fireMatteo Collina · JSConf.Asia 2018
• 26mObserving Node.js: Using Metrics to Improve your Application PerformanceGuilherme Hermeto · OpenJS World 2021
• 46mNodeJS Scalibility TipsLuciano Mammino · CityJS Conference 2020
• 24mArchitect for Scale - My Learnings in Node.js (A case study)Shreyansh Pandey · JSFoo Coimbatore 2019
• 21mCan you tell me if your Node app is healthy?Alejandro Oviedo García · Nordic.js 2018
• 25mNode js applications diagnostics under the hoodNikolay Matvienko · JSConf AU 2018
• 01h 12mA New Way to Profile Async Activity in Node.jsDavid Mark Clements, Mathias Buus · Node+JS Interactive 2018
• 27mScaling NodeJS beyond the ordinaryAbhinav Rastogi · JSConf Iceland 2018