John Allspaw's AMA
summary
Etsy runs roughly 40–60 deploys a day while deliberately decoupling code rollout from schema and infra changes, using Chef to manage the infrastructure substrate rather than as an all-in-one deploy tool. Alert overload drives measurable sleep loss tracked with Jawbone/Fitbit and teams treat alerting as a design problem. A public post criticizes the SEC Knight Capital writeup for explaining rule violations but not the causal story. Focus is on culture, measurement and system safety.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 1:48Would you have expected the current state of the industry
- 4:47Why did you leave Flickr for Etsy
- 7:35Advice for large enterprises
- 10:10Continuous deployment
- 18:20Any story is the story
related talks
32mContinuous deployment at scale
24mContinuous Deployment at Scale
16mHuman Centered On Call Practices
36m"Alert overload: How to adopt a microservices architecture..."
33mDevSecOps: How to Use DevOps to Make You More Secure
47mPersistent SRE Antipatterns: Pitfalls on the Road to Creating a Successful
24mMeatbag Systems: How Our Reliability Culture & Practice...
41mSleeping with one eye open: production support
