You deployed a new application to Google Kubernetes Engine and are experiencing some performance degradation. Your logs are being written to Cloud Logging, and you are using a Prometheus sidecar model

Question

Options

• ACreate custom metrics from the Cloud Logging logs, and use Prometheus to import the results
• BExport the Cloud Logging logs and the Prometheus metrics to Cloud Bigtable. Run a query to join
• CExport the Cloud Logging logs and stream the Prometheus metrics to BigQuery. Run a recurring
• DExport the Prometheus metrics and use Cloud Monitoring to view them as external metrics.

• A
  4% (1)
• B
  4% (1)
• C
  8% (2)
• D
  85% (22)

Explanation

The correct approach is to export Prometheus metrics into Cloud Monitoring as external metrics. Cloud Monitoring natively integrates with both Cloud Logging (for log-based metrics and log correlation) and Prometheus (via the Managed Service for Prometheus or the external metrics API), providing a unified observability platform. It supports real-time alerting policies at no extra query cost beyond ingestion. This minimizes cost compared to exporting to data warehouses. Option A reverses the data flow (logs → Prometheus doesn't make sense for alert correlation). Option B (Cloud Bigtable) is expensive, not designed for ad-hoc analytical queries or real-time alerting, and requires custom join logic. Option C (BigQuery with recurring queries) adds latency and cost-BigQuery is optimized for batch analytics, not real-time alerting, and query costs accumulate with frequent scheduled queries.

No community discussion yet for this question.