The deployment pipeline stages
A deployment pipeline is the automated path a change travels from a commit to running in production. Each stage is a gate. The earlier a problem is caught, the cheaper it is to fix, so the pipeline front loads fast checks.
Typical stages
- Build, compile the code and produce a versioned artifact once, then reuse it everywhere
- Test, run unit and integration tests, plus static analysis, to catch regressions
- Stage, deploy the artifact to an environment that mirrors production for end to end checks
- Release, promote the same artifact to production, often with canary or rolling strategy
A core principle is build once, deploy many. The exact artifact that passed tests is the one that ships, so you never rebuild between stages and risk a different result.
Properties of a good pipeline
- Fast feedback, with quick checks first so failures surface in minutes
- Repeatable and automated, so a deploy is a button, not a ritual
- Promotion gates, where each stage must pass before the next begins
- Traceability, linking a running version back to its commit and tests
Key idea
A pipeline gates a change through build, test, stage, and release, promoting one artifact built once so what ships is exactly what passed.