Docker and the Three Ways of DevOps: System Thinking, Feedback Loops, and Continuous Learning

Gene Kim’s novel The Phoenix Project introduced the four IT work types and the three ways of DevOps, which are now explained in the context of Docker.

The First Way – System Thinking

System thinking means viewing the software change process end‑to‑end, from customer request to production deployment, to avoid local optima and silos. Docker contributes by enabling fast, isolated environments for development, integration, and deployment, such as multi‑container test stacks, Docker‑based CI slaves, and blue‑green deployments.

Increase the speed of each process component.

Reduce resource and time waste in sub‑processes.

Use functional isolation to visualize and understand the global flow.

The Second Way – Amplify Feedback Loops

Feedback loops allow quick detection and correction of problems. Docker images encapsulate both application and infrastructure, eliminating differences between development, testing, and production environments, thus speeding up error identification and reducing lead time.

Rapid error correction is essential.

Reduced variation simplifies problem localization.

Embedding metadata in images (build time, source repo, tags) further accelerates troubleshooting.

The Third Way – Culture of Continuous Experimentation

Organizations practicing the third way foster continuous learning (Kaizen) and treat every change as an experiment (Plan‑Do‑Study‑Act). Docker serves as the “lab equipment” that enables safe, repeatable experiments and immutable delivery pipelines.

Real‑world examples include Gilt’s immutable delivery using Docker, a financial institution’s CaaS platform that cut data‑tool pairing lead time from days to hours, and an R‑container for baseball analytics.

In summary, Docker‑enabled pipelines reduce delivery cost and risk while increasing deployment frequency, embodying the three DevOps ways.