Can your backend really handle 100k concurrent users?
summary
Claims a real end-to-end load test that reaches 100k concurrent users against a Django-postgres-redis-celery on Kubernetes stack, showing which bottlenecks appear: Postgres connection limits and query hotspots, Redis blocking and eviction under pub/sub and cache pressure, Celery backlog growth and worker saturation, and Kubernetes pod autoscaling lag. Provides concrete fixes applied (connection pooling, query indexes and batching, Redis sharding/timeout tuning, trimming task fan‑out, HPA and probe tuning) and reports which changes cut error rates and latency versus which still need more capacity planning.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 4:31Understanding the problem
- 10:04Initial application setup
- 16:28Scaling challenges
- 20:04Advanced optimizations
- 24:33Conclusion and Q&A
related talks
15m"Stress Less: Easy Database Load Testing Using Python and Locust"
36mAutomated Performance Testing of Postgres
21mCrunching Millions of Requests for Breakfast: A Tale of Performance Testing Prowess
31mLoad Testing Kubernetes: How to Optimize Your Cluster Resource Allocation in Production [I]
45mBeat the rush! Designing effective load tests for your web application
35mLoad Testing Made Easy with Gatling
45mSchrödinger's Pokémon: Observability for Chaotic Load Testing
35mStress Testing as a Culture