Abstract: Not everyone is ready to do Continuous Delivery (CD) but there is pretty strong agreement that moving in that direction is a good thing for many (most?) applications. We hear a lot about the important of automation, but this talk focuses on how to think about what you need to automate (hint: it’s not run every test you can cobble together). At various steps in the delivery cycle you need make the decision to either promote a piece of code to the next step or go back and make further changes/fixes. Scaling development multiplies these decision points as integration points in which teams collaborate around shared functionality.
Too many organizations approach CD by automating the wrong things and just increasing the amount of (often poor quality) code they have to maintain. Others focus only on various provisioning/configuration management tools and declare themselves “DevOps enabled” without really improving their ability to deliver small chunks of new code. We’ll discuss what functioning CD really looks like and why it helps the organization so much.
Also, we’ll discuss the requirement for valid test environments in order to make good decisions, and what the tests should look like that enable our decisions. Finally we will talk about how to assess and improve our overall delivery system over time resulting in better software faster without the stress and drama of major releases.
Speaker: Kelly Looney, Skytap