29 Jan 2021 -
The main thrust is for each single code change (pull request, merge request, etc) to be deployed by the author when they feel it is ready. They deploy the change, ensure the deploy succeeds, and watch how it works in production. Then if necessary they only have to roll back their own deploy.
Essentially the idea is to change they deployment strategy from a single bulk operation to a stream of atomic transactions.
Ideally you track:
And you want to target:
Under this model, each developer has full ownership and control over their changes and works in a tight feedback loop to sharpen their skills.