Q94 — AWS DVA-C02 Ch.1
Question 94 of 100 | ← Chapter 1
A company is developing a serverless application composed of various AWS Lambda functions behind an Amazon API Gateway API. Developers want to automate deployment of Lambda function code. They will use AWS CodeDeploy to deploy updated Lambda functions. The deployment must minimize potential errors experienced by end users. While the application is in production, it must not experience downtime outside of scheduled maintenance windows. Which deployment configuration meets these requirements with the shortest deployment time?
- A. Use AWS CodeDeploy in-place deployment configuration for the Lambda function. Immediately shift all traffic after deployment.
- B. Use AWS CodeDeploy linear deployment configuration to shift 10% of traffic per minute.
- C. Use AWS CodeDeploy all-at-once deployment configuration to immediately shift all traffic to the updated version.
- D. Use AWS CodeDeploy predefined Canary deployment configuration to immediately shift 10% of traffic, then shift the remaining traffic after 5 minutes. ✓
Correct Answer: D. Use AWS CodeDeploy predefined Canary deployment configuration to immediately shift 10% of traffic, then shift the remaining traffic after 5 minutes.
Explanation
The predefined Canary deployment configuration helps reduce production disruption risk by gradually introducing traffic to the new version. First, 10% of traffic is immediately shifted to the new version, limiting impact if issues arise. Then, after 5 minutes, the remaining 90% of traffic is shifted, allowing validation of stability and performance before full rollout. This significantly reduces potential errors and downtime for end users during updates. Compared to other options: Option A (in-place deployment) risks impacting all users during deployment. Options B (linear) and C (all-at-once) lack gradual traffic shifting advantages and increase production disruption risk. Therefore, Option D best satisfies the requirements for shortest deployment time while minimizing potential errors.