A Murder of Crows: Actors, Instrumentation, and Load-testing as Part of the TDD Cycle
summary
Claims that load testing belongs inside the TDD cycle and demonstrates it with concrete numbers and tools: C-based routines beat pure Erlang by about 2× on tiny ops and 20–30× on XML parsing, a custom tool amok runs scenarios 5–20× faster than k6, and a large run simulates 100,000 users with messages every 5 ms and 60 messages per session. Tests report ops/sec baselines, latency distributions, SLAs, and automated statistical thresholds so regressions and gradual drift are detected, plus stress tests that intentionally kill the system.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:09Speaker background
- 3:33Why performance matters
- 6:28Performance testing basics
- 10:48Messaging server demo
- 17:14Load testing with A MOK
- 18:38TDD cycle with benchmarks
- 23:12Defining SLAs
- 25:39Measuring performance
- 27:43Handling benchmark noise
- 30:56Choosing load tools
- 33:07A MOK vs K6
- 35:39Advanced load configs
- 37:28Instrumentation and data
- 38:56Key takeaways
- 39:25Q&A
related talks
31mKill! Kill! Die! Die! Load Testing With A Vengeance
45mBeat the rush! Designing effective load tests for your web application
21mCrunching Millions of Requests for Breakfast: A Tale of Performance Testing Prowess
39mLast-Minute Performance Testing
37mExploratory Load-Testing with AI Tooling - it's Time to Ramp Up!
37mRethinking How We Test Our Async Architecture
48mHow to not Crash in Prod, Load Testing Crash Course
01h 10mPerformance Testing Crash Course