Which description matches a canary deployment pattern with progressive traffic slicing?

Which description matches a canary deployment pattern with progressive traffic slicing?

• A. Only test in development environment.
• B. Deploy new version to all users at once.
• C. Gradually increase traffic to the new version while monitoring metrics; stop or rollback if thresholds are breached.
• D. Roll back after one hour with no metrics.

Answer: (C)

The idea being tested is gradually exposing users to a new version, watching how it performs in real conditions, and expanding the rollout only if everything stays within safe limits; otherwise, you stop or roll back. This is exactly what the described choice conveys: progressively increase traffic to the new version while monitoring metrics, and halt or rollback when thresholds are breached. This approach limits risk (blast radius) and provides data to confirm stability before full deployment. In contrast, testing only in a development environment doesn’t involve real production traffic; deploying to all users at once removes the safety net of incremental checks; and rolling back after a fixed time without metrics ignores actual performance signals.

The idea being tested is gradually exposing users to a new version, watching how it performs in real conditions, and expanding the rollout only if everything stays within safe limits; otherwise, you stop or roll back. This is exactly what the described choice conveys: progressively increase traffic to the new version while monitoring metrics, and halt or rollback when thresholds are breached. This approach limits risk (blast radius) and provides data to confirm stability before full deployment. In contrast, testing only in a development environment doesn’t involve real production traffic; deploying to all users at once removes the safety net of incremental checks; and rolling back after a fixed time without metrics ignores actual performance signals.