You operate a microservices application running in AKS with .NET and Node.js services. You need to: 1) Reduce Application Insights ingestion costs while preserving full end-to-end transaction details whenever an operation fails. 2) Test external availability of the public API endpoint from three regions at 1-minute intervals. 3) Automatically detect anomalies like sudden increases in dependency failures. What should you implement?

The correct option addresses all requirements. Instrumenting with the Application Insights SDK and enabling adaptive sampling at the SDK level reduces ingestion costs while preserving full transaction details for failed operations, as adaptive sampling prioritizes retaining correlated telemetry for failures. Creating a Standard (preview/new) availability test from three regions at 1-minute intervals directly fulfills the external availability monitoring requirement. Ensuring Smart Detection is enabled allows for automatic anomaly detection, such as sudden increases in dependency failures, without manual configuration. Incorrect options: Disabling SDK sampling and using only 5% server-side ingestion sampling might lose correlated telemetry for failed transactions. Using Log Analytics for traces and Workbooks for availability is less efficient than dedicated Application Insights features. Enabling Profiler for all services and disabling sampling would be very costly and not suitable for general anomaly detection. Using Traffic Manager health probes for availability is not its primary purpose for detailed application availability, and static sampling at 1% might not preserve correlated telemetry for failures.