DevOps: CI/CD, Infrastructure as Code and GitOps

1. A team releases once a month, and each release bundles hundreds of changes. Releases often fail and take days to debug. Which change most directly reduces the risk of each release?

• Ship smaller sets of changes more often
• Add a longer manual test phase to each release
• Release every quarter so testing has more time
• Freeze all code changes for a week beforehand

2. Management tells every team to reach 20 deployments a week. Teams start splitting trivial edits into many deploys to hit the number. What is happening?

• Deployment frequency is not one of the metrics DORA says to track
• The teams are now top performers and simply need a harder target
• Splitting changes always lowers the change failure rate on its own
• The measure became a target, so it stopped reflecting real improvement

3. A director wants to rank all teams against each other by their DORA numbers and reward the top five. What is the main problem with this?

• DORA metrics can only be measured for teams that already practise fully automatic deployment
• The four metrics are owned by Google, so other companies are not allowed to measure them
• Teams work in different contexts, so the numbers are most useful for tracking each team's own trend
• Ranking only works when every team uses the same programming language and cloud provider