AZ-400 Question 157
Single answerYour team manages a microservice-based e-commerce platform. You want to release new features gradually to specific user segments�initially testing them with only a small subset of premium subscribers�and revert quickly if errors occur. You also want to capture feedback for iterative improvements. Which deployment approach best fits these needs?
- A
Blue-green deployment with environment swap
- B
Canary release to 100% of the user base in a single step
- C
Feature flags combined with ring-based (progressive) deployment
- D
A/B testing with two completely separate environments for all users
Show answer and explanation
Correct answer: C
Explanation
Feature flags and ring-based deployment together provide a powerful strategy for gradually rolling out new features, allowing quick rollback in case of issues, and limiting exposure to specific user segments before the feature is fully deployed. This approach is well-documented in Azure DevOps best practices for progressive exposure. More details can be found in the official Azure DevOps documentation under deployment strategies and feature flags.
- A. Incorrect.
Option 1 (Blue-green deployment) focuses on maintaining two identical production environments where one is 'live' and the other is 'idle' until you switch traffic. While it reduces downtime, it does not address releasing to selective user segments or gradual rollouts in small increments. Hence, it doesn't meet the requirement for targeting small subsets of users.
- B. Incorrect.
Option 2 (Canary release to 100% of the user base in a single step) applies the concept of a canary release but nullifies its benefit by deploying to all users at once. A true canary release gradually ramps traffic from a small group to a larger one, so deploying to the entire user base immediately doesn't fit the risk-mitigation goal described in the scenario.
- C. Correct.
Option 3 (Feature flags combined with ring-based (progressive) deployment) allows for selective exposure of new features to specific subgroups�such as premium users�through feature flags while also introducing changes gradually to different 'rings' or segments of the user population. This approach provides the necessary flexibility to roll back quickly if issues arise and gather feedback from targeted users. This best meets the stated requirements.
- D. Incorrect.
Option 4 (A/B testing with two completely separate environments for all users) is intended for comparing features or versions simultaneously in production. However, A/B testing typically splits users across two or more versions for marketing or user experience experiments. It is not primarily focused on a progressive, safe rollout to specific user segments nor reversion at scale if errors emerge.