Skip to content

Metrics

For device data that is synchronized to the cloud, you can create custom metrics from the data stream. These metrics can then be displayed or graphed in the Device Manager.

Ioto metrics automatically calculate statistics for the last "5 minutes", "hour", "day", "week" "month", or "year". For each timespan, you can query the average, minimum, maximum or sum value for the metric.

Metric Naming

Metrics are created and subsequently queried by a set of naming properties including:

  • namespace
  • metric name
  • dimensions

The namespace is a global name to group related metrics. Currently, this must be set to Embedthis/Device.

The metric name is the specific metric name. For example: a device's temperature.

The dimensions select specific instances of a metric. For example, each device may have a temperature metric and we may also have an overall device pool average temperature metric.

Defining Metrics

In the process section for a model, define a metrics array of metric definitions. Each definition selects a data item value to be converted to a metric.

For example:

const DeviceSchema = {
    process: {
        Fault:   { 
            sync: 'up' 
            metrics: [{
                namespace: 'Embedthis/Device',
                fields: [{Temperature: 'temp'}],
            }]
        }
    },
    ...
}

This will create a temperature metric from the item's temp attribute.

Metric definitions may contain the following properties:

Property Type Description
namespace string Metric namespace for the metric. Must be set to 'Embedthis/Device'.
fields array Array of metrics to create. Array metric entries may contain field attribute names or a map of a field attribute to a metric name.
where string Conditional expression to select items for which to create metrics.
buffer map Metric buffering directives. Defaults to null

The metric namespace scopes the metrics and must be set to 'Embedthis/Device'. Other namespaces may be supported in the future.

The fields property contains an array with one or more metric definitions. Each definition may be either:

  • An item attribute name which is used to identify the value AND to name the metric.
  • A map of an item attribute name to a metric name.

For example:

1
2
3
{
    fields: ['temperature', 'status', 'speed']
}

This would create three metrics using the item's temperature, status and speed attributes. The metrics would take the same name.

Wheras:

1
2
3
{
    fields: [{Temp: 'temperature'}, {Online: 'status'}, {Level: 'speed'}]
}

This would create the metrics: Temp, Online and Level from the temperature, status and speed item attributes.

Where Expressions

The where property can be used to select matching items (rows).

The where query language is based on familiar Javascript expressions with some additional operators. Item attributes are expressed as variable names and literal values are expressed as JavaScript values.

For example: the expression:

1
error == "critical" && component == "PS1"

will select those items which have the error attribute set to "critical" and the component attribute set to "PS1".

The query language supports the operators: + - * / ( ) ^ ! % == != < <= > >= ^= ^!= $= $!= << >> && || <> ><

These extension operators have the following meaning a ^= b a starts with the string b a ^!= b a does not start with the string b a $= b a ends with the string b a $!= b a does not end with the string b a >< b a contains the string b a <> b a does not contain the string b

Sub-expressions can be grouped with parenthesis and the boolean operators && and || can group conditional operands.

Regular expressions (delimited by slashes) may be used with the "==" and "!=" operators. The regular expression can be on either side of the operator.

The query language understands the types: Numbers, Boolean, String literals, Regular Expressions and null.

For example:

{
    ModelName: {
        metrics: [
            {
                namespace: 'Embedthis/Device',
                fields: [{CPU: 'value'}],
                where: 'key == "cpu"',
            }
        ]
    }
}

Displaying Metrics

The Device Manager can display metric data via graphical widgets

Manager

You can select from the following widget types:

  • gauge
  • graph
  • table
  • progress
  • number
  • leds

When creating or modifying a widget, you can select the data source to be either a database table item or a metric.

When selecting a metric, you enter the metric name, statistic and resource dimensions.

When selecting data directly from the database, you enter the database table model name, the model field (attribute) and a row selection expression.

Manager Widget Edit