Components

A metric is defined by three elements: the dimensions that segment it, the aggregations that calculate its values, and the filters that restrict the events considered.

Dimensions

Dimensions are fields of the event type used to segment the metric into distinct categories. Only fields with discrete values can be used.

For example, a metric with the status dimension generates a separate time series for each possible value of that field (approved, declined, pending), instead of a single global count.

A metric can have up to 10 dimensions. The combination of dimensions is defined at creation and cannot be changed afterwards.

A metric without any dimension represents a single time series, with no subdivisions.

Aggregations

An aggregation is the calculation applied to events in each time interval. Each metric must have at least one aggregation, and it is possible to configure several to compare different perspectives on the same chart.

Available aggregation functions:

Function	Description
Count	Number of events in the interval
Sum	Sum of values of a numeric field
Minimum	Lowest value of a numeric field
Maximum	Highest value of a numeric field
Average	Average of values of a numeric field
Percentage	Proportion between two sets of events

Each aggregation can have its own name and its own filters, applied only to it, independently of the metric’s filters.

Aggregations over fields that represent a duration are displayed already converted to a readable unit (for example, 3.5s), both in the table and in the chart.

Metric filters

Filters applied at the metric level restrict the events considered by all aggregations. They are useful for limiting scope from the configuration onwards, without needing to repeat the filter in each aggregation.

The metric filters and the aggregation filters are applied at the moment the data is aggregated, and for that reason they do not accept references to lists. Changes apply to the next aggregations and do not reprocess history already consolidated.