問題文
A team needs to compare error rates between two versions under real traffic, and their versions can coexist. Which approach gives them the comparison while limiting exposure?
選択肢
- A canary release, since a small share of traffic reaches the new version while metrics are evaluated.
- A blue-green switch, since the two environments can be compared before the switch.
- Recording both versions as declared state at the same time and letting the agent converge toward whichever one produces fewer errors, so that the decision is made by the reconciliation loop rather than by a separate analysis.
- A/B testing, since routing by request attribute means the same users always reach the same version and the comparison is therefore between two stable populations rather than between two samples of mixed traffic.