continuous delivery

continuous delivery

What to measure

• outcomes not output
• outcomes not utilization (i.e. velocity, loc, etc)

what to actually measure:

1. delivery lead time
2. deployment frequency (proxy for batch size)
3. time to restore service
4. change failure rate

Principles:

1. Built in quality
2. Small batches
3. Computers perform repeatitive tasks; people solve problems
4. Relentlessly pursue improvement
5. Everyone is responsible

Foundations:

1. Comprehensive configuration management
2. Continuous Integration
3. Continuous Testing

Trunk based development is a direct indicator of performance? I remain dubious about trunk based git flow.

src: Accelerate