We will describe our journey from a process where design/planning work was performed away from development to one where small cross-functional Feature Teams self-organized to complete design, planning, and construction within the same sprint. Each team member is involved in getting READY, planning, executing and being DONE. The results we observed are an increase in team morale, more predictable results and accumulation of less debt, while maintaining a constant velocity. Our process is a deviation from the established approach where upfront work needs to be ready before starting a sprint.
How do you scale Scrum to hundreds of people? This presentation will explain a way of organizing your development so that it scales up well. It involves breaking the link between architecture and organization, breaking code ownership and organize the development in a more customer centric way. This has its drawbacks too! These are explained and some techniques for overcoming these drawbacks are discussed. This talk is based on the “feature teams” and “requirement areas” chapters in the recently published “Scaling Agile & Lean Development” by Bas Vodde and Craig Larman.
In 2005, Microsoft’s DevDiv (with 2000 participants and 40 million lines of code) overhauled its engineering practices to improve agility, quality, and customer satisfaction. Four years into the journey, customer satisfaction has increased dramatically. Product quality improved 10x. Velocity improved 2x, with schedule time for major releases was cut by eighteen months and quarterly releases of “power tools” allowed incremental delivery to external customers. Practices that change include planning, org, quality gates, branching, testing, tooling, reporting, backlogs, transparency.