A development team wants to ship a new checkout flow behind a feature flag so they can enable it for a small percentage of users first and disable it instantly if errors rise, all without redeploying the application. They want the flag change itself validated and rolled out gradually rather than flipped for everyone at once. Which AWS service is purpose-built to manage and safely roll out these feature flags?
- AAWS Secrets Manager, which stores the flag value as a secret and rotates it on a schedule so the checkout flow toggles automatically across the user base.
- BAmazon CloudWatch Evidently alarms, which watch the checkout error metric and flip a stored configuration key back to its previous value when a threshold breaches.
- CAWS CodeDeploy, which performs a canary deployment of the new code so a subset of users receives the checkout flow before the full fleet does.
- DAWS AppConfig, which serves feature flags as managed configuration and supports validators and a gradual deployment strategy with automatic rollback on alarm. Correct
Why A is wrong: Secrets Manager is built for credentials and rotation, not feature flags, and it has no gradual rollout or validation, so it cannot stage the flag change safely.
Why B is wrong: A CloudWatch alarm can detect errors but does not itself store or gradually deploy feature flags, so it cannot manage the flag rollout the team described.
Why C is wrong: CodeDeploy shifts code and traffic, but the team wants to toggle behaviour with no redeploy, so a code deployment tool does not meet the no-redeploy flag requirement.
Why D is correct: AppConfig has a dedicated feature-flag configuration type, runs validators before release, and rolls the flag out gradually with CloudWatch alarm-triggered rollback, matching every requirement.