DevOps Practices (Flow, Feedback, Continual Learning)
Domain 1 - DevOps Practices (Flow, Feedback, Continual Learning)
This domain tests whether you can map practices to the Three Ways and choose the intervention that improves the whole value stream. Expect scenario questions about lead time, batch size, deployment pipelines, telemetry, and learning loops. Questions often ask for the next best action, not the most fashionable tool.
The Three Ways as design rules
- Flow is left-to-right work movement: make work visible, limit WIP, reduce batch size and handoffs, and optimize for global throughput rather than local utilization. Deployment pipelines, trunk-based development, and feature flags support flow.
- Feedback is right-to-left information: automated tests, monitoring, telemetry, canary checks, and Andon-style stop-the-line behavior shorten detection time.
- Continual learning turns experience into capability: experimentation, blameless review, knowledge sharing, and deliberate practice.
Worked example
A team reports a 12-day lead time. Value-stream mapping shows 7 days waiting for a shared test environment, 2 days in test execution, and 3 days in approval. The shared environment is the constraint. Appropriate moves: limit work in progress entering test, split large changes, create on-demand test environments, and automate regression tests. Reasoning: adding testers would improve local test utilization but worsen the queue and global flow. The first intervention targets the constraint; the second shortens feedback; the third captures learning in the pipeline.
Traps that cost marks
- Equating DevOps with tools or a team name.
- Optimizing a busy resource while queue time grows.
- Treating feedback as only postmortems; it also includes tests, telemetry, and peer review.
- Celebrating large batch releases because they look efficient.
- Adopting canary releases without enough telemetry to interpret the result.
In scenario questions, first identify the system constraint, then choose the smallest change that improves flow and feedback without creating new queues. A practice earns marks only when it changes the system behavior the scenario describes.