Load testing distributed web services
summary
loveholidays serves 8,050 requests per second at peak and enforces daily load testing by replaying recorded access logs through Ripley, an open‑source Go tool. logs stream from Engine X into Loki, Albot converts them to jsonlines, and Ripley replays them at multiples (tested up to 40x/80x). initial runs reveal timing drift, fixed by a wall‑clock compensation and HTTP connection tuning; nightly replays catch regressions and proved effective during January peaks after production tuning.
AI-written summary · about
chapters (25)
- 0:00Load testing intro
- 1:05Speaker background
- 3:58Business context
- 5:47Search latency challenges
- 7:56Kubernetes autoscaling
- 9:24Cost vs reliability
- 10:48Why load testing matters
- 11:52Load testing tools limits
- 13:59Architecture and logging
- 15:26Alo load testing tool
- 17:15Ripley traffic replay
- 20:31Ripley profiling
- 23:24Ripley timing drift
- 26:19Ripley in production
- 28:48Ripley HTTP limits
- 30:14Replaying incidents
- 32:02Load testing privacy
- 33:34Fun prop and Go
- 34:40Engineering principles
- 38:17Chaos engineering
- 41:09API failures handling
- 44:22Ripley daily runs
- 46:30Replay risks mitigation
- 48:38Graceful degradation
- 51:51Q&A
related talks
47mLoad testing distributed web services
28mAccelerate with confidence through AWS Distributed Load Testing
47mLoad Testing You Always Wanted
33mHold My Beer - Load Testing. In Production. On Autopilot.
26mLoad testing websites following the steps of a user journey
26mIntroduction to Load Testing
30mBeat the rush! Designing effective load tests for your web application
26mTo see what others can not... You must climb the mountain, A Introduction to Load Testing