Shipping frequently and safely.
WHAT CONTAINERS CONTRIBUTE
One artefact, identical from testing to production.
WHY THAT MATTERS
It removes the largest source of deployment surprises.
WHAT THE PIPELINE SHOULD PRODUCE
An image, tagged with the commit, tested and scanned.
WHAT DEPLOYMENT SHOULD BE
Updating a reference to that image.
WHY THAT IS THE WHOLE IDEA
Deployment becomes a configuration change rather than a build.
WHAT TO NEVER DO
Build a different image for production.
WHY
The tested artefact is then not the deployed one.
WHAT TO PROMOTE BETWEEN ENVIRONMENTS
The same image, with different configuration.
WHAT TO AUTOMATE
Testing Building Deployment to non-production
WHAT TO GATE
Production, on whatever approval your context requires.
WHAT TO MEASURE
Time from commit to production Deployment frequency How often deployments fail How long recovery takes
WHY THOSE FOUR
They describe delivery capability better than anything else.
WHAT TO IMPROVE FIRST
Recovery time.
WHY
It determines how much risk each deployment carries.
WHAT SMALLER DEPLOYMENTS PROVIDE
Less risk each, and easier diagnosis.
WHAT TO AIM FOR
Deployments boring enough to happen daily.
