Hey, psst!? Are you Continuously Deploying yet? You know everybody is doing it, right?

Why?

Some to go faster. Some to increase quality. Some because, let’s face it, it’s a really cool buzzword.

But let’s talk about what the real reason is. Should be. The control necessary to do Hypothesis Driven Development. Or, as I like to call it: Test First Product Management.

With the popularity of Lean Startup, we’ve seen a significant uptake in these kinds of business experiments. But how does it work in practice? What tools can we use to get started? Should we be building some new ones to help out?

In this talk, I’ll go through some examples, showing how to build up a system that allows you to start with HDD right now. And how to structure your process around that along the way.

  • What is HDD?
    • Theory, Lean Startup, Scientific Method, TDD everything
    • Format
    • Decide up front what to do for each possible result
  • Process
    • From Feature Backlog to Experiment Generation
    • Business Goals as Roadmap
      • Aside: Alignment
      • Aside: Scaling
    • Demo business results
  • Implementation
    • First things first: getting data
      • Keep It Simple (aka: you already have it!)
      • Analytics, ELK, etc.
    • Transparency: Dashboarding is Easy
      • From docs to dashing
    • Feature Toggles and A/B tests
      • Basics (the 10 minute solution)
      • Slightly less basic: canaries, automation
  • Where we’re going:
    • New tooling for expressing experiments
    • (Semi-) Automated dashboarding
    • Integration into the Delivery Pipeline

 


by Wouter Lagerweij

Session: October, 3nd – DevOps Forum

Organization

Institutional Partners