A production incident is in progress. You need to watch near real-time request rate, failure rate, and dependency calls for a single App Service instance in the production slot without waiting for data to be stored. You also want to filter to only failed requests and dependencies to reduce noise. What should you use?

Application Insights Live Metrics Stream provides near real-time (sub-second latency) telemetry directly from your running application, making it ideal for monitoring during an active incident without waiting for data ingestion. The server-side filtering allows you to specify the production slot and a particular instance, and also filter for failed requests and dependencies, reducing noise and focusing on critical issues. Metrics Explorer provides aggregated metrics, not real-time streaming, and while it has dimension filters, it wouldn't offer the immediate, live view needed. Application Insights Profiler is for detailed performance analysis, not real-time operational monitoring, and focuses on code execution paths rather than request/failure rates. A Log Analytics query would operate on stored data, introducing latency, and wouldn't provide the immediate, streaming view required for a live incident.