Which solution will meet these requirements?
Create a canary release deployment stage for API Gateway. Deploy the latest API version. Point an appropriate percentage of traffic to the canary stage. After API verification, promote the canary stage to the production stage.
Create a new API Gateway endpoint with a new version of the API in OpenAPI YAML file format. Use the import-to-update operation in merge mode into the API in API Gateway. Deploy the new version of the API to the production stage.
Create a new API Gateway endpoint with a new version of the API in OpenAPI JSON file format. Use the import-to-update operation in overwrite mode into the API in API Gateway. Deploy the new version of the API to the production stage.
Create a new API Gateway endpoint with new versions of the API definitions. Create a custom domain name for the new API Gateway API. Point the Route 53 alias record to the new API Gateway API custom domain name.
Explanations:
A canary release allows for gradual traffic shifting to the new version of the API, minimizing the impact on users. By verifying the new API version’s performance with a small percentage of traffic, issues can be detected early, and if everything works well, the canary stage can be promoted to production with minimal disruption and data loss.
This option suggests creating a new API endpoint and importing the new version in merge mode. However, this approach can cause disruptions if there are changes that are not compatible with existing integrations. Merging can also lead to complexities in API behavior, which may not be optimal for a smooth transition.
Similar to option B, this option involves creating a new API endpoint and using the overwrite mode. While overwrite mode replaces the existing API, it does not provide a mechanism for gradual rollout or verification of the new API version, increasing the risk of downtime or data loss during deployment.
Creating a completely new API endpoint and changing the Route 53 alias record can lead to significant downtime for users and potential data loss during the tr