Load-test a Cloud Run–based discussion portal composed of microservices (some calling databases) and identify per-request bottlenecks following Google best practices.

Cloud Trace is the most effective tool for identifying per-request bottlenecks in a microservices architecture like the Cloud Run-based discussion portal. It provides detailed, end-to-end distributed tracing, allowing you to visualize the latency of each service call within a request. This directly pinpoints which specific microservice or database operation is causing delays. While emitting per-request elapsed time series to Cloud Logging and using Log Analytics heatmaps can show overall latency distribution, it doesn't provide the granular, per-request breakdown across services that Cloud Trace offers. Exposing latency metrics for Prometheus is good for aggregate monitoring but lacks the distributed tracing capability to see the flow and timing of a single request across multiple services. Adding timing log statements and analyzing them in BigQuery is a manual and less efficient approach compared to the automated distributed tracing provided by Cloud Trace.