Knobs, buttons & switches: Operating your application at scale
summary
Concrete levers and how they’re implemented matter: use maintenance mode as an env var and a read‑only mode, implement feature flags per group (backed by Postgres or Redis), apply rate limits with a global default and per‑account modifier (example default 100 req/min that can be halved to shed load), stop noncritical jobs via a report setting, wrap risky changes with scientist gem, and add responsive and manual circuit breakers. Build a single dashboard and run game days to verify these controls actually work.
AI-written summary · about
chapters (5)
related talks
24mKnobs, Levers and Buttons: tools for operating your application at scale
22mKeep Ops Happy, Designing for Production
40mGrowing internal tooling from the console up
01h 02mTools for simplified large scale systems
27mScaling Ruby Applications: Challenges, Solutions and Best Practices
35mFeature Toggles: The Good, The Bad, and The Ugly
27mTradeoffs on the road to Observability
27mIt's Down! Simulating Incidents in Production