Step 01 of 06
Understand the idea
Deploying by hand makes it easy to miss a step. A delivery pipeline runs the steps in a repeatable order and records what changed, which checks passed, and which version was deployed.
Key technical ideas
CI answers whether the change is safe to merge
Continuous integration (CI) automatically checks code changes before they are accepted. The checks in this course format and analyze the code, run tests, check for data races, and build the program. A passing run shows that these checks succeeded; it does not prove the code has no bugs.
Delivery prepares a releasable artifact
Continuous delivery means each accepted change is prepared so it can be released. The pipeline builds, scans, signs, and stores the software. A person or policy may still need to approve putting it into production. The same prepared artifact should be used for that release.
Deployment changes what users run
Continuous deployment automatically releases a change to users after it passes the required checks. Continuous delivery prepares a change for release but may wait for approval. Teams choose continuous deployment when they can monitor changes and recover safely.
DORA metrics keep the argument honest
DORA metrics help teams understand how quickly they deliver changes and how safely they do it. Lead time and deployment frequency measure delivery speed. Change failure rate and time to restore measure how often releases cause problems and how long recovery takes. Use the measures together.
stages:
- test
go-test:
stage: test
image: golang:1.25
script:
- go test ./...
rules:
- if: '$CI_PIPELINE_SOURCE == "push"'
- if: '$CI_PIPELINE_SOURCE == "merge_request_event"'Save this as .gitlab-ci.yml in the project root. GitLab checks out the code, starts a Go container, and runs the commands under script. The rules start the job for pushes and merge requests.
CI, delivery and deployment are not synonyms
| Practice | Trigger | Ends when | Risk moved |
|---|---|---|---|
Continuous integration | Push or merge request | Checks pass | Bad code is stopped early |
Continuous delivery | Merge or release request | Artifact is releasable | Release is a decision |
Continuous deployment | Passing gate | Users run it | Rollback and observability matter |
Manual deploy | Someone remembers | Commands finish | Human memory carries risk |
Technical terms used at work
Trust boundary — The line between source someone wrote and an artifact allowed to affect users.
Lead time — Elapsed time from committed change to running in an environment.
Change failure rate — The share of deployments that cause incidents, rollbacks or urgent fixes.
Time to restore — How long the team needs to return service after a bad change.
