Continuous Deployment at Scale
At Etsy our approach to development is to make small, incremental, continuous changes to the site. Deploying code to production anywhere between 20 to 40 times a day requires not only technical tooling, but also a culture that allows for and encourages innovation, confidence backed by actual data and comprehensive tests, and blameless postmortems that allow for a feedback loop we can learn from and improve.
Learn about the tools we have, and how we foster an environment where we make this possible.
Outline/structure of the Session
Continuous Deployment Principles
- core principles
- traditional deployment cycle
- Frequent check-ins
- Feature flags
- A/B testing and experimentation
- dark changes
- config pushes
- Individual tooling
- push train and IRC
- post-deploy tooling
- learning from failure
Our understanding of how failures happen is fundamentally broken, which leads to an a culture without trust. Tooling alone cannot help our continuous delivery goals - culture plays a critical role in both enabling and in being a critical feedback loop into improving our systems. In this talk we will learn how Etsy achieves continuous deployment through both technical tooling and fostering a culture that allows for it.
Engineers, Engineering Managers, Program/Project Managers