Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
This article describes the range of detection settings available when you create a rule. You learn how detection settings operate in Fabric Activator and how to configure them effectively. The various detection settings work together to pinpoint the exact data that you're interested in tracking.
The examples in this article use the Package delivery events sample eventstream. It specifically references the Temperature attribute and Too hot for medicine rule created in the tutorial.
Detection setting options
You configure detection settings in the Definition pane, which appears on the right side of Activator. The pane opens when you select a rule in the Explorer pane on the left. The detection settings in Summarization, Condition, and Property filter work together to define exactly which data events trigger the rule. Each setting is described in the sections that follow.
The following image shows the Definition pane with the detection settings for the Temperature attribute of the Too hot for medicine rule.
Summarization
A summarization converts a stream of raw events into a single computed value over a rolling time window. It's made up of an aggregation type (Operation), a Window size, and a Step size. Summarizations are useful when a rule should respond to a trend or pattern over time, rather than to individual events.
The Window size defines how far back in time each computation looks. For example, a window size of 1 hour means each summary value is computed from events in the most recent hour. The Step size controls how frequently the window advances and a new summary value is computed. For example, a step size of 15 minutes means a new aggregated value is produced every 15 minutes. Both values can range from 10 seconds to 24 hours.
The following table describes the available aggregation types.
| Aggregation | Description |
|---|---|
| Average | Computes the average value of the property or column over the time window. |
| Minimum | Computes the minimum value of the property or column over the time window. |
| Maximum | Computes the maximum value of the property or column over the time window. |
| Sum | Computes the sum of the property or column values over the time window. |
| Total | Computes the number of events containing the property or column over the time window. Unlike Sum, which adds up the values of the property, Total counts how many events occurred. |
Conditions
A condition defines the pattern that Activator must detect in the data to activate the rule. Conditions are grouped by data type, and each group contains condition types that describe different kinds of change or state.
The following image and table describe the condition categories available in the drop-down.
| Category | Description |
|---|---|
| Numeric change | Conditions that activate when a numeric value changes relative to a threshold, such as Increases above or Decreases below. Use these conditions to detect directional trends in numeric data. |
| Numeric state | Conditions that activate when a numeric value is in a particular state, such as Is greater than, Is less than, or Is between. The rule activates for each event where the condition is true. |
| Text change | Conditions that activate when a text value changes to or from a specific value, such as Changes to or Changes from. |
| Text state | Conditions that activate when a text value matches a pattern, such as Contains, Begins with, or Ends with. |
| Logical change | Conditions that activate when a boolean value changes state. Becomes true activates when a value changes from false to true. Becomes false activates when a value changes from true to false. |
| Logical state | Conditions that activate for each event where a boolean value matches the specified state. Is equal to and Is not equal to compare the value against true or false. |
| Common change | Activates when an attribute value changes. The Changes condition applies across data types and has no specific threshold or target value. |
| Heartbeat | Conditions based on event arrival. No presence of data activates when no new events arrive within a specified time. Object first appearance activates the first time a given object ID appears in the event stream. |
Occurrence
For some condition types, an Occurrence field appears that controls how long, or how many times, the condition must be true before the rule activates. For example, if you want to be alerted only when a temperature remains above 100 degrees for at least 10 minutes, set the condition to Is greater than 100 and set the occurrence to When it has been true for 10 minutes. The following table describes the available occurrence options.
| Occurrence | Description |
|---|---|
| Every time the condition is met | The rule activates each time the condition is true. |
| When it has been true for n times | The rule activates only after the condition is true the specified number of times. |
| When it has been true for | The rule activates only if the condition remains continuously true for the specified duration. |
Property filter
The Property filter section creates a condition on a subset of your data, limiting which events the rule evaluates. Activator supports property filters on any type of attribute - numeric, text, or boolean.
Each filter specifies an Attribute, an Operation, and a Value. Only events that satisfy the filter are passed to the rule condition. All other events are excluded.
For example, the image shows a filter on the Temperature (°C) attribute where Operation is Is greater than and Value is 0. This filter excludes all events where Temperature is zero or below. Another example is a text filter where Attribute is set to City, Operation is Is equal to, and Value is Redmond - limiting rule evaluation to events where packages are in Redmond.
You can apply up to three filters to a single rule. When you apply multiple filters, an event must satisfy all filters for the event to be evaluated. Filters are combined with AND logic.
Advanced settings
The Advanced settings section of the Definition pane contains timing settings that affect rule evaluation accuracy, not detection logic. The Wait time for late-arriving events setting controls how long Activator holds the evaluation window open to allow delayed events to arrive. To learn more, see Latency in Activator.