CD Pipelines
A Continuous Delivery/Deployment pipeline automates the journey of code from version control to production. The goal is reliable, repeatable, and fast releases.
Pipeline Stages
┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
│ Source │─▶│ Build │─▶│ Test │─▶│ Staging │─▶│Production│
│ │ │ │ │ │ │ │ │ │
│ Git push │ │ Compile │ │ Unit │ │ Deploy │ │ Deploy │
│ │ │ Package │ │ Integration│ │ Verify │ │ Monitor │
│ │ │ │ │ Security │ │ Accept. │ │ Rollback │
└──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘
Deployment Strategies
Rolling Update:
v1: [●][●][●][●][●] → v2: [○][○][○][○][○]
One by one, zero downtime
Blue-Green:
Blue: [●][●][●][●][●] ← live traffic
Green: [○][○][○][○][○] ← deploy new version
Switch traffic when ready
Canary:
v1: [●][●][●][●][●] ← 90% traffic
v2: [○] ← 10% traffic (canary)
Gradually increase if healthy
A/B Testing:
Route users based on headers/cookies
Compare metrics between versions
Pipeline Best Practices
- Fail fast — Run cheapest tests first, stop on failure
- Parallel stages — Run independent jobs simultaneously
- Environment promotion — Same artifact across dev, staging, prod
- Rollback capability — Every deployment should be reversible
- Feature flags — Decouple deployment from release