AZ-400 Question 38
Single answerYou are part of a DevOps team managing a microservices-based e-commerce application hosted in Azure App Service and Azure Kubernetes Service (AKS). Users have reported intermittent slow checkouts at unpredictable times. You need to detect the root cause of these performance issues by implementing the most effective approach to gather and query operational metrics. Which approach should you take?
- A
Enable standard Azure Monitor metrics on each resource without customizing or correlating the data.
- B
Use Azure Monitor Logs with custom KQL queries and log-based metrics to correlate request durations, exceptions, and resource utilization across services.
- C
Rely on CPU utilization alerts in the AKS cluster to determine when performance issues occur.
- D
Collect diagnostic logs only for Azure App Service without including AKS cluster metrics.
Show answer and explanation
Correct answer: B
Explanation
To diagnose intermittent performance problems in a microservices architecture, you need to correlate logs and metrics from all critical components. Azure Monitor Logs and custom KQL queries enable a holistic view across services, making it possible to detect root causes of slowdowns. Refer to Microsoft documentation on Azure Monitor (https://learn.microsoft.com/azure/azure-monitor/) for best practices on log correlation and creating log-based metrics.
- A. Incorrect.
Incorrect. Turning on standard metrics without correlation will provide some visibility, but it will not help identify how multiple services interact or where exactly the bottleneck occurs.
- B. Correct.
Correct. Using Azure Monitor Logs and custom KQL queries allows you to correlate multiple streams of data, such as request durations, exceptions, and resource consumption. Creating log-based metrics helps you pinpoint the source of intermittent slowdowns across microservices.
- C. Incorrect.
Incorrect. While CPU usage can be a useful metric, relying solely on CPU alerts does not capture network latency, memory constraints, or application-level exceptions. This approach misses deeper insights into request-level performance issues.
- D. Incorrect.
Incorrect. Limiting data collection to just the App Service ignores possible problems originating in other services running on AKS. A comprehensive approach requires telemetry from all relevant resources.