Abstract
Uber's automated API management platform provides an ability for developers to ship quickly, automating API's writing, testing and production. I'll cover lessons learnt from migrating 1500 internal API's to a next-generation system - it's like changing the tires of a moving F1 car. Growing and scaling engineering organizations is hard on developer productivity. In this talk, I'll focus on the key learnings from building the next generation API Management Platform at Uber. In it's first-generation, API management was a single repository with 1000s of engineers making code changes to interact with 3000 internal microservices. This led to bottlenecks in code-reviews, bugs which affected trips globally and caused large scale outages. In this talk, we will review the lessons learned from rebuilding this first-generation system into a fully automated API Management Platform in GoLang. The automated API management platform provides an ability for developers to ship quickly, with automation handling everything needed to write an API, testing and making it to production. I'll cover the lessons from a large scale migration - migrating 1500 API's internally to a new system. We will talk about centralized vs decentralized migration techniques, what works and what does not. It's like changing the tires of a moving F1 car.
