Explore Tab

The Explore tab helps you identify which label values contribute most to a metric's cardinality. While other views show the total number of time series or label combinations, this tab allows you to drill down further and see how individual label values affect the number of generated series.

This helps you move from "this metric has high cardinality" to "these specific label values are driving it." By understanding which values generate the most series, you can decide whether to adjust label usage, remove unnecessary dimensions, or change metric design.

How to use the Explore tab

The Explore tab works in three steps.

Step 1: Select a label

Select a label associated with the metric. The label selector lists all labels detected for the metric. Once selected, the interface displays the available values for that label.

Step 2: Select label values

Select one or more label values to analyze. The values panel displays the values associated with the selected label. You can select up to 40 values at a time. This allows you to compare how different values contribute to the total number of time series.

Step 3: Review the cardinality table

The table displays the results for each selected value, showing:

• Value: The label value
• Series: The number of unique time series generated
• Samples: The number of samples associated with those series

Use this table to compare values and identify which ones generate the most time series. This helps determine whether specific label values are responsible for cardinality growth and where to focus optimization efforts.

Time selection

The Explore tab operates on a 1-hour time window. Use the time selector to choose the hour you want to analyze. Only one hour can be selected at a time.

Advantage over PromQL

The same cardinality information can be queried using plain PromQL, but PromQL queries are subject to the series scan limit. If a metric has very high cardinality, a PromQL query may fail or return incomplete results once the limit is reached.

The Explore tab overcomes this limitation. It can scan any number of series without hitting any scan limit for the selected hour, giving you a complete and accurate view of cardinality regardless of metric size.

Common use cases

Assuming the appropriate labels are present on the metric, use the Explore tab to answer questions such as:

• Which region produces the most cardinality? Select the region label to compare series counts across regions and identify which ones generate the most time series.
• Which cluster produces the most cardinality? Select the cluster label to find clusters with disproportionately high series counts that may need optimization.
• Which service produces the most cardinality? Select the service or service_name label to pinpoint services driving cardinality growth.