whenever anyone performs deployment; focus on the metrics associated with the feature. ii. If negative results, ideally feature toggle off and fix-forward or use other Blue- Green, Canary release patterns Engineering) iii. Find the proper balance between fixing production issues and new features development – feature is only “done” when it’s performing as expected in production c. HAVE DEVELOPERS FOLLOW WORK DOWNSTREAM actually exists.” ii. How to build a feature: 1. Ask, “Should we build it, and why?” 2. Perform cheapest, fastest experiment possible to validate whether the feature will achieve desired outcomes. 3.

as we decide whether the quality of each individual feature is adequate for the feature to be deployed.  It is easy enough to say that the feature is ready for deployment when it passes all of its tests concepts of deployability and quality. A feature is simply deployable when it passes its tests; that is a yes-or-no question independent of quality.  Once the feature is in production, we can assess its

when validation tests fail 3. Developers working in small batches on trunk rather than long-lived feature branches f. BUILD A FAST AND RELIABLE AUTOMATED VALIDATION TEST SUITE i. Slow and periodic feedback purposes ii. Deployment – installing a specified version to a given environmentiii. Release – when a feature is made available to customers. 1. Release of functionality should not require changing our application Pattern – modify application to selectively release functionality by small configuration changes. Feature flags, dark launching h. ENVIRONMENT-BASED RELEASE PATTERNS i. Enables deployments during normal

After we have invested so much of our time into implementing a given feature, if our customer tells us they don’t value the feature, can we really let it go?

day- or week- long improvement blitzes with Dev & Ops teams. Focus on problems they care about. NO FEATURE WORK ii. Goal is focused improvement on daily work, not experimentation and innovation iii. Demo