Configure GKE workload metrics using kubectl. Select all Pods to send their metrics to Cloud Monitoring. Create a custom dashboard of application metrics in Cloud Monitoring to determine performance bottlenecks of your GKE cluster.

Update your microservices to log HTTP request methods and URL paths to STDOUT. Use the logs router to send container logs to Cloud Logging. Create filters in Cloud Logging to evaluate the latency of user requests across different methods and URL paths.

Instrument your microservices by installing the OpenTelemetry tracing package. Update your application code to send traces to Trace for inspection and analysis. Create an analysis report on Trace to analyze user requests.

Install tcpdump on your GKE nodes. Run tcpdump to capture network traffic over an extended period of time to collect data. Analyze the data files using Wireshark to determine the cause of high latency.