AZ-400 Question 209
Single answerYou manage a set of microservices that each trigger separate CI pipelines in Azure DevOps. Over the past few weeks, your organization� pipeline concurrency usage has notably increased, driving up costs without providing consistent improvements in build times. You need to balance concurrency to optimize performance while controlling expenses. What should you do to achieve this goal?
- A
Configure a manual approval step in each pipeline just before deployment, ensuring no pipeline runs without explicit consent
- B
Implement concurrency limits in pipeline settings so only a certain number of pipelines can run in parallel, with excess pipelines queued
- C
Require all pipeline runs to happen after working hours to minimize usage during the business day
- D
Use a single monolithic pipeline for all microservices to reduce the total number of CI runs
Show answer and explanation
Correct answer: B
Explanation
To optimize pipeline concurrency for both performance and cost, you can configure concurrency limits in Azure DevOps. This tactic ensures that only a defined number of pipelines run simultaneously and others remain queued, preventing sudden spikes in build agents and associated costs. According to Microsoft� Azure DevOps documentation, controlling concurrency is a recommended approach to balance build frequency with budget constraints. By applying concurrency limits appropriately, you gain more granular control over pipeline parallelism while still allowing frequent, reliable builds.
- A. Incorrect.
Incorrect. While adding manual approvals can reduce concurrency, it slows down development workflows unnecessarily and doesn�t systematically address parallel build usage or costs.
- B. Correct.
Correct. By limiting concurrency at the pipeline level, Azure DevOps queues additional runs instead of starting them all at once. This mitigates cost spikes by controlling how many builds run simultaneously and still keeps pipelines running at a manageable pace.
- C. Incorrect.
Incorrect. Scheduling pipelines to run only after hours might reduce peak concurrency cost temporarily, but it isn�t practical for teams needing continuous feedback during the workday. It also introduces delays in development.
- D. Incorrect.
Incorrect. Combining everything into a single large pipeline can complicate and slow down the build process if any single microservice has issues. It doesn�t directly address cost optimization related to concurrency settings.