Designing a CI/CD Pipeline

A CI/CD pipeline automates the path from code commit to production deployment.

Pipeline Stages

Code Push
    ↓
CI Pipeline:
  1. Checkout code
  2. Install dependencies
  3. Lint + static analysis
  4. Unit tests
  5. Integration tests
  6. Build Docker image
  7. Push to registry
    ↓
CD Pipeline:
  8. Deploy to staging
  9. Smoke tests
  10. Manual approval (production)
  11. Deploy to production (canary → full)
  12. Post-deploy tests

Test Pyramid

       E2E Tests (slow, few)
      Integration Tests
    Unit Tests (fast, many)

Aim: 70% unit, 20% integration, 10% E2E.

Docker Build Optimization

# Layer caching — dependencies before code
COPY package.json .
RUN npm install          # Cached if package.json unchanged
COPY src/ .              # Only this invalidates cache
RUN npm run build

Multi-stage build:

FROM node:18 AS builder
RUN npm run build

FROM node:18-alpine AS runtime
COPY --from=builder /app/dist .
# Final image: ~50MB instead of ~500MB

Artifact Management

Docker Registry: ECR, GCR, Docker Hub
Helm Charts:     Git repo, ChartMuseum
Maven/npm:       Nexus, Artifactory

Environment Strategy

Dev     → Auto-deploy on every commit
Staging → Auto-deploy on merge to main
Prod    → Manual trigger (or canary auto)

Rollback

Auto rollback: Error rate > 5% in 10 min
Manual:        Redeploy previous image tag

Tools

Category	Tools
CI	GitHub Actions, Jenkins, GitLab CI
CD	ArgoCD, Spinnaker, Flux
Registry	ECR, GCR, Harbor
K8s deploy	Helm, Kustomize

Conclusion

Great CI/CD pipelines run in <15 minutes, enforce test gates, and enable one-click rollbacks. Speed and safety are equally important.