Docs » Charts » Specifying Chart Options in the Chart Builder

Specifying Chart Options in the Chart Builder


Need some context? Plotting Metrics and Events in the Chart Builder


In addition to customizing individual plots on a chart, you can set a number of options on the Chart Options tab that apply to the entire chart.

The options that are available depend on the type of chart; no chart type supports all the available options. All possible options are listed alphabetically below.

Color by dimension

Color by dimension is generally appropriate when you have one metric that you’re looking at across multiple sources, and want to be able to compare what those sources are doing. For example, in a chart that is displaying API latency by availability zone, coloring by dimension helps you compare latency across zones.

Color by metric

Color by metric is generally appropriate when you have more than one metric you are looking at in a given chart. For example, if you are displaying cache hits in plot A and cache misses in plot B, coloring by metric lets you compare the total number of hits with the total number of misses.

You can also use plot names to ensure that plots representing similar metrics and dimensions are displayed in different colors. For the purpose of plot display color selection, a different plot name is interpreted as a different metric. If you want two plots with similar signals and dimensions to appear in different colors on the chart, edit the plot names to make sure they contain different text, and select Color by metric. After you have done so, the colors of the plots will be different from one another.

Color by value

  • On a single value chart or list chart, you can use this option in conjunction with its secondary visualization type to have the colors on the chart represent the status of the metric, based on thresholds you specify. For example, if a value goes above (or below) a threshold, the number can be displayed in red. This lets you see status at a glance when looking at the chart on a dashboard.

    Note that using Color by value overrides any plot color setting you might have specified in the plot configuration panel.

  • On a heatmap chart, you can use the Fixed option for Color thresholds to specify threshold ranges and colors. These values determine what colors are used for squares in the chart.

A similar option is available for histogram charts (see Color theme).

Using with a secondary visualization option

Important

If you are using the Fixed color threshold option on a heatmap chart, this section doesn’t apply; skip to Specifying threshold ranges and values.

  1. Select the secondary visualization type you want to use. The illustrations in this section show a single-value chart with a secondary visualization of Radial.

  2. If necessary, Select Value from the Color by dropdown menu to display the threshold selector. (If you have already specified a secondary visualization of Radial or Linear, Color by value is the only option available.)

    ../_images/config-color-options-value-01.png
  3. If you are using a secondary visualization of Radial or Linear, set minimum and maximum values (or accept the defaults). These values let SignalFx know how to display the lower and upper boundaries of the visualization.

    ../_images/min-max.png

Continue to the next section to set ranges and colors.

Specifying threshold ranges and values

When specifying threshold ranges, you start by entering a single value; by default, numbers above this value will be displayed in red and numbers below or equal to the value will be displayed in green. (You can change these colors, as shown in step 3 below.)

You can specify up to four values, which can define up to five color ranges. To specify if a range should be defined as greater than or equal to a value (>=), as opposed to the default of greater than a value (>), click the > symbol to the left of the value.

  1. Enter the first (highest) value you want to represent in the range. Numbers must be in order from highest to lowest. For example, if you plan to have values of 25, 50, and 75, enter 75 first. (Numbers must be in descending order, so it is easiest to enter them in that order. However, you can edit them at any time, as long as they are in descending order when you are finished.)

    ../_images/config-75.png
  2. Click + to increase the number of color ranges. If you change your mind about the number of ranges you want to specify, hover over the value you want to remove and click the x that is displayed to the right of the value.

    ../_images/config-threshold-delete-value.png
  3. By default, SignalFx assumes that low values are desirable (green) and high values are undesirable (red), which is appropriate for metrics such as Latency or CPU Utilization. To set thresholds where lower values are undesirable (for example, if the metric is something like Available Memory), simply click on the color chips to change them to your desired color. You can use one of the standard colors (shown below) or click More to see a color palette with more colors to choose from.

    ../_images/high-is-good.png

As you enter values for ranges, the color changes based on the thresholds you entered. For single-value charts, a border in the same color is added. In the illustration below, the number is yellow and there is a yellow border because the value is in the range between 25 and 50.

../_images/config-color-options-value-02.png

On a dashboard, the border lets you determine at a glance if you have used Color by value to specify thresholds. This feature is especially useful when you use a secondary visualization of Sparkline or None with a single-value chart, because you are not seeing the threshold ranges as you are with Radial or Linear visualizations.

In the illustration below, the border on the left indicates that value is orange because it meets a threshold condition. The color of the value on the right simply reflects the color that was set (or is the default) for the plot in the chart.

../_images/config-threshold-borders.png

Color theme

Specifies the color family to use when populating a histogram chart (see Graph charts). The color you select will represent the darkest value on the chart; other values will be shown with progressively less saturation.

Similar options are available for heatmap charts (see Color thresholds) and for single-value and list charts (see Color by value).

Color thresholds

Specifies whether squares on a heatmap chart should be colored from light to dark in a single color range (see Auto color threshold) or should be colored based on color ranges and values you choose (see Fixed color threshold).

Similar options are available for histogram charts (see Color theme) and for single-value and list charts (see Color by value).

Auto color threshold

By default, heatmap charts have a Color threshold setting of Auto, with no minimum or maximum values specified. This means that:

  • Squares are colored from light to dark in a single color range.
  • Each color represents one of 5 ranges, based on the actual minimum and maximum values at the time the chart is refreshed (based on its resolution setting). For example, if values range from 0 to 100, the lightest squares represent values ranging from 0 to 20 and the darkest represent values ranging from 80 to 100.
  • There are always 5 ranges, but you might not see all ranges represented on the heatmap if you have no sources reporting a value in a given range.
  • Square shading is dynamic, and can change as the minimum or maximum value changes.

You can customize the auto threshold coloring in the following ways:

  • Specify a fixed minimum value, a fixed maximum value, or both.
For example, suppose you know that most of the values in your chart will be between 0 and 1000, with a few outliers in the range of 5000. If you don’t set a maximum, the outlier values will be taken into account when shading the squares, which will give you a less representative display. Instead, set a maximum value of 1000; the bulk of your squares will be shaded in 5 ranges between 0 and 1000, and any values over 1000 will be displayed in the darkest color, regardless of their actual value.
  • Choose a different color scheme.
The default color scheme is shades of green. Click one of the color swatches next to the Max or Min fields to choose a different color scheme, or greyscale.

Fixed color threshold

Choose Fixed from the Color thresholds dropdown menu to display the threshold selector, which lets you specify how many color ranges you want to display and the values that each range reflects. The colors of the squares update dynamically based on their values and the ranges you have specified.

../_images/heatmap-threshold.png

For example, suppose the squares represent percent of cache misses per host. If you want all hosts reporting values higher than 30% to be colored red, choose Fixed and set a single threshold value of 30. Hosts with cache misses below 30% will appear green, and those above 30% will appear red.

For more information, see Specifying threshold ranges and values.

Data table columns

Specifies which columns you want to display in the data table.

By default, all dimensions relevant to the plots on the chart are displayed, along with one or more other fields. To specify which fields are displayed, click Custom. Toggle items on and off as desired.

../_images/data-table-columns.png

Tip

For information on editing the plot names that are displayed, see Plot name.

To re-order the fields, click and drag the icon that appears at right when you hover over the items on the list.

../_images/columns-reorder.png

Default time

The default time range applied to most new charts is the last 15 minutes (-15m). However, if a new chart contains AWS-specific metrics, the default time range is the last hour (-1h). This is because AWS metrics are reported less frequently than most other metrics, so a range of one hour is more likely to contain a useful number of datapoints to display.

Depending on the purpose of the chart, you might want to see values for a longer or shorter time period. Use this option to change the default time range for a chart. The value you specify will be applied whenever you open the chart or view it in a dashboard, unless there is a time range override.

Time range and heatmap charts

By default, a heatmap chart reflects the datapoint received when the chart last refreshed; charts refresh every 5 minutes. You can specify an absolute time range to see values representing the last datapoint received at an earlier time. For example, if it is now 3 PM, you could specify a time range ending at 1 PM to see what the heatmap values were approximately 2 hours ago.

Tip

If you want to see past values, be sure not to choose a relative time range from the Time Range Selector; this will just continue to display the most recently received datapoint. Instead, you must specify an absolute time range.

Description

In addition to providing a title for a chart, it’s often a good idea to provide additional information about the chart. Providing this information helps other users in your organization understand the data being displayed in the chart.

Display fields

Specifies which fields you want to display alongside list values in a list chart. For more information, see Data table columns.

Disable sampling

In cases where a large number of time series would be displayed, e.g. if you choose a metric being reported by 500 servers, SignalFx samples a subset of those time series so the chart will render more quickly. The sampled display provides you with an approximate sense of the values in those time series. Analytics still apply to all data. If you disable sampling, any time series data that were previously omitted will be shown. Depending on the number of time series, disabling sampling may cause the chart to render more slowly.

When data is being sampled, you’ll see a message like this on the chart:

../_images/sampling.png

If you click Click here to disable sampling, or check the Disable sampling checkbox in the chart options tab, the sampling message is no longer displayed, and any time series data that were previously omitted will be shown. Depending on the number of time series, disabling sampling may cause the chart to render more slowly.

Group by

Lets you select up to two levels of grouping for your data. In the following illustration, results are grouped by plugin_instance within aws_availability_zone.

../_images/heatmap-group-by.png

Include zero on Y-Axis

When selected, ensures that a value of zero is included on a Y-axis that is being dynamically scaled to accommodate data values.

When plotting values on a chart, SignalFx by default dynamically scales the Y-axis so that the minimum and maximum values are close to the lowest and highest values of the signal. For example, if values range between 2 and 5, the lowest value on the Y-axis will be approximately 2. Similarly, if the values range between -5 and -2, the highest value on the Y-axis will be approximately -2.

Instead, you might want the respective minimum or maximum displayed to be zero. Including zero can give you a sense for the scale of the values, as well as for the absolute size of the changes or fluctuations over time.

In the following illustration, the chart on the right has this option enabled.

../_images/include-zero.png

If you have specified a minimum or maximum value for an axis (see Using the Axes tab), zero might not be shown on the Y-axis even if this option is enabled. For example, if you set a minimum value of 50 or a maximum value of -20, zero will not be shown on the Y-axis; conflicting minimum or maximum values on an axis override this option.

Max delay

SignalFx sets the Max Delay parameter based on estimates of how ‘on time’ the time series are. By default, the Auto setting lets SignalFx detect and apply a reasonable value automatically, based on how your data is coming in.

If you know that some of your data is being delayed, and you don’t want to wait for that data to arrive before your charts are updated, then you can set max delay accordingly. To ensure that charts are updated as quickly as possible (i.e. don’t wait for late data at all) select 1s for max delay.

To specify your own values, enter the number in milliseconds; e.g. enter 2000 to specify a max delay of 2 seconds. The upper limit is 15 minutes (900000 ms) but values over 5 minutes are generally not recommended. If you enter a value higher than 15 minutes, max delay will be set to 15 minutes. For more information, see Delayed datapoints.

The value you specify will be applied whenever you open the chart or view it in a dashboard, unless there is a max delay override.

Maximum precision

Specifies the number of digits to display for a value on a single value chart or list chart. When precision is Automatic (the default), the number of digits displayed depends on the space available. The examples shown below compare results when using the values of 2, 3, or 4, but other values are also acceptable. The actual number of digits displayed may be more than the maximum you specify, depending on the value; for example, whole numbers are always displayed in full.

Value Maximum precision Display
1235.76 2 1236
  3 1236
  4 1236
     
23.576 2 24
  3 23.6
  4 23.58
     
0.23532 2 0.24
  3 0.235
  4 0.2353

Minimum resolution

Specifies the minimum interval for which SignalFx should roll up values to display a datapoint on the chart. For example, if you are tracking the number of support calls received per hour, you might not want to see a chart that shows datapoints representing the number of calls received every 15 minutes, even if data is available at that resolution. Setting this option to 1h ensures that the datapoints will always represent values for periods of 1h or more.

Order by

Specifies how you want the squares on a heatmap to be sorted. For example, if you want to see the largest or smallest values in a predictable location in the heat map, select Value. You can sort by any dimension or property associated with the metric. Click the small arrow to the right of the Order by option to toggle between ascending and descending sort order.

Note that on a list chart, a sorting option is also available, but it is called Sort instead of Order by.

Refresh every

While graph charts refresh in real time, some other chart types (such as single value or list charts) refresh only periodically. For these charts, you can specify a “Refresh Every” setting in the Chart Options tab to specify how frequently to update the display. (Of course, in practice the refresh interval cannot be lower than the data resolution.)

This behavior can have strange side effects when paired with the lag that can be observed for incoming data, as is sometimes the case with AWS CloudWatch data. For example, if a list is displaying the current value for a subset of incoming time series (e.g. latency for some number of ELB load balancers), e.g. Top 25, and the time series are reporting at a 5m resolution, but the refresh interval is set to 5s or 1m, then chances are good that at any particular refresh not all of the time series will have reported, and the list will appear more sparsely populated as a result.

Secondary visualization

On a single value chart or list chart, you can use this option to specify how you want the value or list to be displayed.

../_images/seco-vis.png

Sparkline

A sparkline provides a visual representation of how a value has been changing over time. When using this visualization, you can color by dimension, metric, or value.

On a single value chart, the sparkline is displayed below the value. On a list chart, it is displayed to the left of the value.

../_images/spark.png

Radial

A radial secondary visualization displays values in a format that resembles a speedometer. When you select this option, the display is dark grey until you enter at least one value (see Color by value). Radial visualizations are always colored by value.

On a single-value chart, the graphic representation is displayed above the value. On a list chart, the graphic is displayed to the left of the value. On both chart types, the number is displayed in the color corresponding to its threshold range.

../_images/radial.png

Linear

A linear secondary visualization displays values in a horizontal bar. When you select this option, the display is dark grey until you enter at least one value (see Color by value). Linear visualizations are always colored by value.

On a single-value chart, the graphic representation is displayed below the value. On a list chart, the graphic is displayed to the left of the value. On both chart types, the number is displayed in the color corresponding to its threshold range.

../_images/linear.png

None

On a single-value chart, a secondary visualization of None displays only the value as a large number, with no sparkline or any other graphic representation. On a list chart, values on the list are displayed with no graphic to the left of the numbers. When using this visualization, you can color by dimension, metric, or value.

../_images/none.png

Show data markers

Specifies whether small dots are displayed on the chart, indicating the times at which there are datapoints.

../_images/data-markers.png

Show events as lines

Specifies whether vertical lines are displayed at times where event markers are shown on a chart.

../_images/custom-event-02.png

Show on-chart legend

Lets you specify a dimension to be displayed in a legend below the chart. The legend shows the value of the specified dimension associated with each plot in the chart, in the same color as the plot.

If the chart uses left and right Y-axes, information will be displayed on the left or right side of the chart, according to the axis used by the specified plot.

Show timestamp

Specifies whether to show a time stamp at the bottom of the chart.

Sort

Specifies the order in which entries should be displayed on a list chart.

Note that on a heatmap chart, a sorting option is also available, but it is called Order by instead of Sort.

Stack chart

Specifies whether to stack areas or columns vertically instead of side by side. All plots should use the same Y-axis (see Left and right Y-axes).

../_images/stack.png

You can change the order of the plots to control how the values are displayed in the stack.

Title

The title is displayed at top left in the Chart Builder and is also shown when viewing the chart on the dashboard. You can use the chart: prefix to search for a title when using the global search.

It’s always a good idea to give a chart a short descriptive title. To provide additional details that are visible when the chart is open in the Chart Builder, add a description.

Use IEC units

Specifies whether Y-axis values are shown in decimal units (1k = 1000) or IEC units (1k = 1024).

Visualization type

See Choosing a chart type.