Engineering Without a Safety Net: Where It Works and Where It Hurts
summary
Engineers weigh four concrete pillars — testing, metrics, observability, resilience — and the talk shows how shortcuts play out: a bad YAML deploy takes down South Korea and Japan for 45 minutes, a streaming platform pushes 1.3 billion bits to Tokyo Olympics with minimal automated tests, and products with tens of millions of users often run with only ~20% of these pillars. Practical tools include Dora metrics (deploy frequency, lead time, MTTR, change failure rate), retries/backoff, circuit breakers, and duplicate implementations for feature flags to keep rollbacks safe.
AI-written summary · about
chapters (10)
- 0:00Introduction and background
- 1:04Real-world failure story
- 3:34Cost of bypassing safety nets
- 6:41Different testing approaches
- 8:47Pillars of confidence in engineering
- 13:21DORA metrics for performance
- 17:40Language of confidence in engineering
- 19:47Signs of crumbling confidence
- 21:10Where to prioritize safety nets
- 24:00Conclusion and Q&A
related talks
16mEngineering Failure-Resilient Systems
27mOptimizing the engineering value stream
52mPerformance & Stability Anti patterns
43m3 Ways You’re Screwing up Platform Engineering – And How to Fix It
48mChaos Engineering: The Path to Reliability
01h 02mPractical Chaos Engineering breaking things on purpose to make them more resilient against failure
48mMove fast Build things… Safely!
47mHow to win in software development with tighter feedback loops