Managed Service for Prometheus uses the intelligent detector to identify abnormal fluctuations in KPI time-series data, enabling informed decisions on alerting, automatic data loss prevention, and root cause analysis.
Limitations
This feature is not available for V2 instances.
Detect abnormal data fluctuations
-
Log on to the Cloud Monitor console.
-
In the navigation pane on the left, choose to open the instance list for Managed Service for Prometheus.
-
Find the Prometheus instance and click the link in the Grafana Workspace column.
-
In the left-side navigation pane, click the
icon. In the upper-left corner, select a data source from the drop-down list next to Explore. -
In the text box next to the Metrics browser drop-down list, enter the following sample PromQL statement to detect anomalous data fluctuations for a metric over a specified time range.
anomal_detect (arms_cms_collector_duration_seconds[180m],3)Note-
arms_cms_collector_duration_seconds: the metric name. Replace it with your target metric. -
The PromQL query must return a range vector. Append a time selector (default: [180m]) after the metric name. The second parameter defaults to 3. If you apply aggregate functions first, change the time selector to [180m:] to convert the result to a range vector. Example: anonym_detect (sum(node_memory_free_bytes)[180m:],3).
-