Crunching Millions of Requests for Breakfast: A Tale of Performance Testing Prowess
summary
Describes a performance stack built around bespoke Kafka load generator plus k6 and JMeter, integrated into CI and run on Google Cloud. Shows real numbers: typical targets of 500–1,000 requests/sec with hotspots hitting 2,000 req/sec, stress runs that take hours or overnight, and volume tests that expose missing DB indexes and slow queries. Notes practical failures: background workers skewing results, InfluxDB integration pain, memory costs of some tools, and using k6 scripts for realistic async load and metrics.
AI-written summary · about
chapters (15)
- 0:00Introduction and speaker intro
- 1:53System architecture overview
- 3:04Types of performance testing
- 5:11Environment and dependency issues
- 7:02Testing tools selection
- 9:12Tool performance trade-offs
- 11:25Testing tool comparison
- 12:57Testing infrastructure and workflow
- 14:02Comparing test results over time
- 15:14Q&A session starts
- 15:37Why multiple testing tools
- 16:20Volume testing approach
- 17:51Test execution duration
- 19:03Running tests in CI/CD
- 20:08Load test scale and limits
related talks
40mPerformance Testing and Big Data
25mGame time: A playbook to (unsuccessfully) 10x in a week and (successfully) 10x in a year
45mDistributed load testing with k6
48mPerformance Testing Workshop with k6
01h 10mPerformance Testing Crash Course
34mPerformance testing: from zero to hero with k6 & Azure
16mPerformance testing – under pressure!
14mBreaking the build with K6 load tests