Testing at scale
summary
Practical checklist for testing at scale: build pipelines that run on thousands of nodes (he cites 2,000), weigh test cost vs risk (80% of features are rarely used), and prefer real production components over mocks when latency and licences allow — never mock if you can. Use Terraform/testcontainers, AutoFixture, mutation testing, TLA+, and anomaly-detection on metrics. For ML use golden-input full-pipeline tests (training can take weeks), feature toggles for instant rollback and replay production logs after redaction. Examples include a $440M outage from insufficient integration testing and cosmic‑ray bit flips.
AI-written summary · about
chapters (3)
related talks
16mChallenges of Testing at Scale | Experiments and Experiences
17mContinuous Testing at Scale
01h 02mThe Art & Joy of Testing in Production
26mThe Impact of Testing on a DevOps Pipeline
13mStrategies for Application Testing at Scale in Enterprise
22mTesting in Production
32mTesting in Production
57mAdvanced Testing Techniques: Tips from the trenches