AZ-400 Question 292
Single answerYou are a DevOps engineer for a company running a container-based application on Azure Kubernetes Service (AKS). The application is instrumented with Azure Application Insights and customers have reported intermittent performance degradation, specifically spikes in response times. You need to identify the root cause of these spikes by analyzing the available metrics. Which approach should you take to effectively isolate the cause of the performance issues?
- A
Use the Metrics Explorer in Application Insights to correlate CPU usage, memory consumption, and response times over the same time window.
- B
Enable container health checks in AKS and rely solely on the pod restart count to assess performance issues.
- C
Export the raw log data from Application Insights to a file and filter manually for any errors or warnings.
- D
Set up alerts on CPU usage at the node pool level, assuming that high CPU is always the cause of degraded performance.
Show answer and explanation
Correct answer: A
Explanation
Best practices recommend using Azure Application Insights and Azure Monitor together to correlate multiple types of metrics�such as CPU usage, memory consumption, and request performance metrics�on a shared timeline. This correlation reveals high-level trends and specific bottlenecks, which is more effective than focusing narrowly on a single metric or relying on manual log reviews. Refer to Microsoft� documentation on Azure Monitor and Application Insights (https://docs.microsoft.com/azure/azure-monitor/ and https://docs.microsoft.com/azure/azure-monitor/app/app-insights-overview) for additional guidance.
- A. Correct.
Correct. By using the Metrics Explorer in Application Insights, you can correlate multiple telemetry signals (e.g., CPU usage, memory consumption, response times) on a single timeline or chart. This helps you pinpoint if resource constraints (CPU or memory) coincide with the performance spikes, enabling a more accurate root cause analysis.
- B. Incorrect.
Incorrect. Although container health checks and pod restart counts are useful for basic diagnostics, they do not provide the in-depth correlation needed to isolate intermittent performance degradations. Pods may remain healthy from AKS� perspective while still exhibiting performance issues.
- C. Incorrect.
Incorrect. While reviewing raw logs can be helpful for detailed troubleshooting, relying solely on manual file filtering is inefficient and prone to human error. It lacks the aggregated view and correlation capabilities offered by Azure Monitor and Application Insights.
- D. Incorrect.
Incorrect. Alerts on CPU usage alone might help if CPU saturation is the main factor, but performance issues could also result from memory constraints, external service latencies, or other bottlenecks. Focusing exclusively on CPU usage can lead to missing key insights.