A gradient-descent approach to minimizing software risk, seen in the wild.
- used DataF, initially
- use microservices because they wanted to get away from a single point of failure
- but, few, limited standards
- moved away from using shared libraries because of version hell
- you break the compiler because now everyone is using different languages
- no consistent model of software correctness.. software evolves
- what risks does this manage? Why did they focus on these risks?
- production risk
- therefore, release (very) often
is DevOps maturity measured in releases per day?
- all the tools replace other, existing tools, but are worse because fewer people use them and so the feedback loop is less.