Skip to content

distribution.md

Latest commit

 

History

History
87 lines (59 loc) · 3.55 KB

distribution.md

File metadata and controls

87 lines (59 loc) · 3.55 KB
title Distribution Widget
widget_type distribution
description Graph a metric distribution aggregated across one or several tags.
aliases
/graphing/widgets/distribution/
further_reading
link tag text
/metrics/distributions/
Documentation
Distributions
link tag text
/dashboards/graphing_json/
Documentation
Building Dashboards using JSON
link tag text
/dashboards/graphing_json/widget_json/
Documentation
Widget JSON schema
link tag text
/dashboards/graphing_json/request_json/
Documentation
Request JSON schema
link tag text
/dashboards/querying/
Documentation
Querying

The Distribution visualization shows data aggregated across one or several tags, such as hosts. Unlike the heatmap, a distribution graph's x-axis is quantity rather than time.

This visualization displays only a single query; additional queries are disregarded.

Note: Outlier detection cannot be performed for this visualization.

{{< img src="/dashboards/widgets/distribution/distribution_fullscreen.png" alt="Distribution graph for JVM heap average by host">}}

Setup

Configuration

  1. Choose the data to graph. The Distribution visualization supports many sources, including metrics, live processes, APM request latency, log events, and RUM events.
  2. Make a selection in the avg/last/max/min/sum control to see your data across the associated tags.
  3. Customize your graph with the available options.

Options

Histogram of points vs. histogram of groups

The Distribution widget supports these histogram types:

  • Histogram of points: Each individual datapoint (for example, a raw request latency measurement) contributes to a bucket based on its value. Use this when you want to see the full distribution of measurements across all recordings.
  • Histogram of groups: Each group (such as a service or host) is first reduced to a single aggregated value (for example, the p75 latency), and then those aggregated values are distributed across buckets. The resulting shape reflects the spread of per-group summary values, not individual measurements.
Because these two types aggregate data differently, switching between them can significantly change the shape of the histogram.

Percentile markers

With APM request distributions, you can add percentile markers on the x-axis.

{{< img src="dashboards/widgets/options/distribution_marker_controls.jpg" alt="Marker control preferences" style="width:80%;">}}

X-axis and y-axis controls

Axis controls are available through the UI and the JSON editor.

They allow you to:

  • Clip the x and y-axes to specific ranges.
  • Automatically change x-axis bounds based on a percentile or an absolute value threshold. This threshold can be applied to one or both ends of the graph (lower and upper) to remove "outlier" bins.
  • Change the y-axis scale from linear to log.

{{< img src="dashboards/widgets/options/distribution_axis_controls.jpg" alt="Distribution axis control preferences" style="width:80%;">}}

Full screen

In addition to the standard full screen options, you can use x-axis controls to zoom in to a specific percentile.

API

This widget can be used with the Dashboards API. See the following table for the widget JSON schema definition:

{{< dashboards-widgets-api >}}

Further Reading

{{< partial name="whats-next/whats-next.html" >}}