You want an alert when the 99th percentile (p99) application latency across backend instances exceeds 200 ms for at least 10 minutes. Which approach will meet this requirement in Google Cloud?

The correct approach is to create a Cloud Monitoring alerting policy. Cloud Monitoring allows you to define alerts based on various metrics, including application latency. The key here is using the p99 (99th percentile) aggregator, which directly addresses the requirement for monitoring the 99th percentile of latency. Setting the condition duration to 10 minutes ensures the alert triggers only when the threshold is exceeded for a sustained period, preventing transient spikes from generating unnecessary alerts. Incorrect options: An uptime check monitors availability, not application latency percentiles. A logs-based metric could track latency, but alerting on the "average" over a 1-minute window doesn't meet the "p99" and "10 minutes" requirements. Cloud Trace is for detailed performance analysis and debugging individual requests, not for aggregate, sustained latency alerting.