AWS DevOps Release Strategies 5 — Questions and Answers
Question 1: A DevOps team wants to ensure that a new microservice release does not degrade the p99 latency of dependent services. Which release strategy combined with which AWS tool best addresses this?
- All-at-once deployment with AWS CloudTrail audit logs
- Canary deployment with Amazon CloudWatch metric alarms and automatic CodeDeploy rollback (Correct answer)
- Blue/green deployment with AWS Config rules for compliance checking
- Rolling deployment with AWS Trusted Advisor performance checks
Correct answer: Canary deployment with Amazon CloudWatch metric alarms and automatic CodeDeploy rollback
A canary deployment exposes the new version to limited traffic while CloudWatch alarms on p99 latency metrics trigger automatic CodeDeploy rollback if thresholds are breached.
Question 2: What is the primary risk of using a 'rolling with additional batch' deployment policy in Elastic Beanstalk compared to a standard 'rolling' policy?
- It is slower because it requires manual approval between batches
- It temporarily runs more instances than the desired capacity, increasing cost (Correct answer)
- It does not support health checks during the deployment process
- It requires a load balancer, adding networking complexity
Correct answer: It temporarily runs more instances than the desired capacity, increasing cost
Rolling with additional batch launches extra instances to maintain full capacity during the update, which temporarily increases infrastructure cost compared to standard rolling.
Question 3: During a CodeDeploy deployment using the 'CodeDeployDefault.ECSLinear10PercentEvery1Minutes' configuration, traffic fails health checks after the first shift. What happens next?
- CodeDeploy pauses and waits for manual intervention
- CodeDeploy automatically rolls back to the original task set (Correct answer)
- CodeDeploy continues shifting traffic at the defined interval
- CodeDeploy sends an SNS notification and terminates the new task set
Correct answer: CodeDeploy automatically rolls back to the original task set
When health checks fail during an ECS linear deployment, CodeDeploy automatically triggers a rollback, restoring all traffic to the original (blue) task set.
Question 4: A team maintains separate AWS accounts for dev, staging, and production. Which AWS service enables automated, policy-controlled promotion of CloudFormation stacks across these accounts as part of a release pipeline?
- AWS Service Catalog
- AWS CloudFormation StackSets with AWS Organizations (Correct answer)
- AWS Control Tower account vending
- AWS Config with multi-account aggregator
Correct answer: AWS CloudFormation StackSets with AWS Organizations
CloudFormation StackSets with AWS Organizations allow centrally deploying and updating stacks across multiple accounts and regions with delegated administration and deployment policies.
Question 5: Which metric is MOST important to monitor during the bake period of a canary release for a database-backed web application?
- EC2 CPU utilization on the canary instances
- Application-level error rate and database query latency from the canary group (Correct answer)
- Total number of HTTP requests routed to the canary group
- CloudFormation stack drift between blue and green environments
Correct answer: Application-level error rate and database query latency from the canary group
Monitoring error rates and database query latency from the canary group reveals application regressions and data-layer performance issues before they affect all users.
Question 6: In AWS CodePipeline, what is the role of an 'artifact' between pipeline stages?
- A CloudWatch metric that tracks pipeline execution time
- An immutable package (stored in S3) that represents the output of one stage and serves as the input to the next (Correct answer)
- A CodeBuild environment variable passed between build and deploy stages
- An IAM role that grants cross-stage permissions within a pipeline
Correct answer: An immutable package (stored in S3) that represents the output of one stage and serves as the input to the next
Artifacts are versioned files stored in S3 that carry build outputs, test results, or configuration from one pipeline stage to subsequent stages.
Question 7: A company wants to minimize the time their production environment runs two versions simultaneously during a blue/green release. Which approach achieves this?
- Use a longer bake period to ensure stability before cutover
- Automate traffic cutover as soon as canary health checks pass, then terminate the blue environment promptly (Correct answer)
- Keep both environments running for 48 hours after cutover to ensure rollback capability
- Use Route 53 with a TTL of 300 seconds to slow down DNS propagation
Correct answer: Automate traffic cutover as soon as canary health checks pass, then terminate the blue environment promptly
Automating the full cutover immediately after health checks pass and terminating the old environment promptly reduces dual-environment running time and associated costs.
AWS DevOps Release Strategies 5
A DevOps team wants to ensure that a new microservice release does not degrade the p99 latency of dependent services.
Which release strategy combined with which AWS tool best addresses this?