Arrested Development: surviving the awkward adolescence of a microservices-based application
summary
Yelp extracts a million-line Python monolith into dozens of microservices and finds coupling, API complexity, and test/deploy pain increase instead of disappearing. The team writes explicit Swagger specs, generates clients with bravado and servers with pyramid swagger, logs everything into Elasticsearch and Cabana, and runs alerts with a last alert. That tooling cuts incident diagnosis from days to minutes, forces intentional interfaces, and makes cross-language clients and API versioning tractable.
AI-written summary · about
chapters (8)
- 0:00Introduction to Yelp and team
- 3:43The case for microservices
- 5:58Challenges with service adoption
- 11:25Case study on service decoupling
- 20:10Defining explicit interfaces
- 26:51Production monitoring and logs
- 33:46Lessons and best practices
- 36:26Q&A session
related talks
33mArrested Development
44mMicroservices: 5 things I wish I'd known
47mBreak Up the Monolith: Testing Microservices
43mBreak Up the Monolith: Testing Microservices
21mThe great SOA migration
01h 00mService Oriented Architecture for Robust & Scalable Systems
37mAPI Horror Stories from an Unnamed Coworking Company
01h 01mMicroservices Gone Wrong
