Break up Monoliths and lay them to REST!
summary
Company runs 75,000 tps at peak and refactors a ten‑year validation monolith that does 18 million validations/day by first strangling it: extract a parser component with its own DB, add APIs, route traffic through an API gateway and feature flags, and shift consumers gradually. Expect doubled infra and temporary workarounds during transition, enforce API contracts and governance, secure expanded surface area, and optionally build a parallel cloud product covering ~80% of use cases for a one‑day cutover.
AI-written summary · about
chapters (18)
- 0:00Intro
- 1:50Welcome
- 2:51About Amadeus
- 3:21Why are APIs important
- 4:02Enterprise API evolution
- 5:53Monolith definition
- 7:22Benefits of Monolith
- 8:40Challenges of Monolith
- 10:48Breaking Monoliths
- 13:13Idealistic Approach
- 14:36Use Cases
- 15:55The Process
- 17:16Validation Engine
- 20:38Separate Component
- 21:48Split Database
- 22:18Switch to backend
- 23:30Autonomous service
- 25:09Parallel tool
