A 45-minute pipeline kills developer productivity. Target: under 15 minutes for the main feedback loop.

Investigation and optimization steps:

┌─────────────────────────────────────────────────┐
│         Pipeline Duration Breakdown              │
│                                                  │
│  Checkout     ██ 2 min                           │
│  Dependencies ████████ 8 min  ← cache this       │
│  Build        ████ 4 min      ← cache layers     │
│  Unit Tests   ████████ 8 min  ← parallelize      │
│  Integ Tests  ████████████████ 15 min ← parallel │
│  Security     ████ 4 min                         │
│  Deploy       ████ 4 min                         │
│               ─────────────────────              │
│               Total: 45 min                      │
└─────────────────────────────────────────────────┘

Optimization strategies:

Cache dependencies — don't re-download npm/pip packages on every build. Cache directories between runs.
Parallelize tests — split unit tests and integration tests across multiple runners
Cache Docker layers — use build cache or remote layer caching so unchanged layers aren't rebuilt
Run only affected tests — if only service-a/ changed, don't test service-b/
Fail fast — run linting and fast unit tests first. Kill the pipeline early if they fail.
Move slow checks to async — full security scans and load tests can run in parallel after deploy to staging
Optimize Docker builds — order Dockerfile layers so dependencies (rarely change) are before source code (changes often)
Use faster runners — bigger machines or spot instances for CI runners
Incremental builds — compile only changed modules, not the entire project

Included with DevOps/SRE Interview Guide

Ready for the complete DevOps/SRE Interview Guide?