Control Prometheus metrics usage
If you’re using a Grafana Agent or local Prometheus datasource to ship your metrics to Grafana Cloud, those metrics arrive as Prometheus metrics. Your active series (the data points derived from your metrics) and the rate at which they are ingested (Data Points per Minute), determine your billable metrics usage. Learn more about how Active Series and DPM affects your billable metrics usage, and how we bill based on the 95th percentile of that usage.
Note: Running and monitoring Kubernetes? Our Grafana Cloud-native Kubernetes Monitoring app is pre-configured and optimized to reduce high cardinality and unnecessary metrics generated by Kubernetes at install. Learn more.
Since metrics usage is based on both active series and DPM, we recommend that you address both of these variables by analyzing and optimizing your Agent’s
remote_write configuration in order to reduce your metrics series ingestion and corresponding usage costs.
1. Optimize your metrics ingestion rate (DPM)
We recommend that the global
scrape_interval in your Agent (or Prometheus) config is set to 1-minute (
60s), to avoid costly high resolution data, by following the guidelines in this article: Optimize your scrape interval to improve data points per minute (DPM)
2. Cardinality Management: Drop unused metrics
Visit the Cardinality Management dashboards (located in the “GrafanaCloud” folder in your Grafana instance) to identify metrics that are not being utilized in dashboards, alerts, or queries. These metrics will be marked
unused. Learn more about unused metrics and how to drop them from your
remote_write configuration in order to reduce your active series usage and associated costs.
3. Cardinality Management: Drop labels for your highest cardinality metrics
From the same Cardinality Management dashboards, analyze those metrics which are generating the highest cardinality. If the metrics are useful and
used, you can still drill down to find the labels generating the highest cardinality for those metrics. Learn how to drop Prometheus labels from those high cardinality metrics in order to reduce your active series usage and associate costs.
Requesting Adaptive Metrics cost optimization tool
Grafana Cloud’s new Adaptive Metrics cost optimization tool is now in public preview! Adaptive Metrics helps you eliminate unused metrics via custom optimizations and recommendations that adapt to your cloud native environment.
Learn more and request access to Adaptive Metrics. Visit our article on how to manually drop metrics with Adaptive Metrics.