How should a solutions architect improve the CI’CD pipeline to reduce the likelihood that changes in the templates will cause downtime?

A company is using AWS CodePipeline for the CI/CO of an application to an Amazon EC2 Auto Scaling group. All AWS resources are defined in AWS CloudFormation templates. The application artifacts are stored in an Amazon S3 bucket and deployed to the Auto Scaling group using instance user data scripts. As the application has become more complex, recent resource changes in the Cloud Formation templates have caused unplanned downtime.

How should a solutions architect improve the CI’CD pipeline to reduce the likelihood that changes in the templates will cause downtime?
A . Adapt the deployment scripts to detect and report CloudFormation error conditions when performing deployments. Write test plans for a testing team to execute in a non-production environment before approving the change for production.
B. Implement automated testing using AWS CodeBuild in a test environment. Use CloudFormation change sets to evaluate changes before deployment. Use AWS CodeDeploy to leverage blue/green deployment patterns to allow evaluations and the ability to revert changes, if needed.
C. Use plugins for the integrated development environment (IDE) to check the templates for errors, and use the AWS CLI to validate that the templates are correct. Adapt the deployment code to check for error conditions and generate notifications on errors. Deploy to a test environment and execute a manual test plan before approving the change for production.
D. Use AWS CodeDeploy and a blue/green deployment pattern with CloudFormation to replace the user data deployment scripts. Have the operators log in to running instances and go through a manual test plan to verify the application is running as expected.

View Answer

Answer: B

Explanation:

https://aws.amazon.com/blogs/devops/performing-bluegreen-deployments-with-aws-codedeploy-and-auto-scaling-groups/

When one adopts go infrastructure as code, we need to test the infrastructure code as well via automated testing, and revert to original if things are not performing correctly.