From 0 $100 billion: Inside lessons on how to scale your dev workflow
summary
Processes over $100 billion a year and explains the concrete practices that scale it: trunk‑based Git with a branch‑per‑release workflow, enforced four‑eyes commits to avoid zombie bugs, and a cultural rule to commit early commit often to prevent merge hell. Uses Postgres as primary DB (50+ TB), central feature toggles with ownership and expiry, streaming for accounting via Kafka patterns but a custom Postgres‑backed streaming queue for exactly‑once consistency, round‑robin DNS for simple multi‑DC failover, releases every three weeks.
AI-written summary · about
chapters (24)
- 0:00Intro
- 1:20What does IDM do
- 2:12Why is IDM so successful
- 3:11Who do we do this for
- 4:07What does that look like
- 4:52Zero downtime
- 5:29Success and failure
- 6:01Maturity vs complexity
- 9:54Being hyperrational
- 14:20Merge hell
- 16:28Release procedure
- 18:19Commit early
- 20:18Encourage communication
- 21:29Zombie bugs
- 23:14Release process
- 26:06Software toggles
- 30:40Avoid early choices
- 37:32Questions
- 38:55Summary
- 40:40Compliance
- 42:52monolithic application
- 43:52version matching
- 45:47Streaming
- 50:12Challenges
related talks
20mLessons Learned from Building a Deployment Process for 100+ Devs at a Large Bank
28mFrom Startup to Scale Up: How to Scale Engineering teams
36mUnlocking the Benefits of Database DevOps and Automation
48mDevops@Scale
27mThe (big) picture of debt
53mVery Large Development: How to Run Code Review for 1000 Developers
29mScaling Across the Enterprise
42mA Modern Development Culture for a Modern Development Process.