Azure Stream Analytics - Windowing Functions

1. Tumbling Window

• Splits data into distinct, non-overlapping, adjoining time intervals.

• An event can belong to only one tumbling window.

• Useful for operations where data in each window is independent.
  sql

• TumblingWindow(timeUnit, windowSize)

  • timeUnit: Unit of time (e.g., second, minute, hour).

  • windowSize: The size of the window in the specified time unit.

  • EXAMPLE: GROUP BY TumblingWindow(second,10)

2. Hopping Window

   • Moves forward in fixed time increments, potentially smaller than the window size, resulting in overlapping windows.

   • An event can belong to multiple hopping windows.

   • Can be used to get frequent insights while still considering a broader time frame.

   • HoppingWindow(timeUnit, windowSize, hopSize)

     • timeUnit: Unit of time.

     • windowSize: The size of the window.

     • hopSize: The time period by which the window moves forward.

     • EXAMPLE: GROUP BY HoppingWindow(second,10,5)

3. Sliding Window

   • Produces an output every time the content of the window changes, i.e., when an event enters or exits the window.

   • Ensures that every output window has at least one event.
     sql

   • SlidingWindow(timeUnit, windowSize)

     • timeUnit: Unit of time.

     • windowSize: The size of the window.

     • EXAMPLE: GROUP BY SlidingWindow(second,10)

4. Session Window

   • Groups events that are close in time while filtering out periods with no data.

   • Session starts with the first event and extends with consecutive events if they are within a specified timeout.
     sql

   • SessionWindow(timeUnit, timeout, maxDuration)

     • timeUnit: Unit of time.

     • timeout: Time after which, if no new data arrives, the session is considered closed.

     • maxDuration: Maximum length of a session.

     • EXAMPLE: GROUP BY SessionWindow(second,5,10)

5. Snapshot Window

   • Groups events with the same timestamp.

   • There's no specific function. Instead, you use System.Timestamp() in the GROUP BY clause.

     • EXAMPLE: GROUP BY System.Timestamp()

source: https://learn.microsoft.com/en-us/azure/stream-analytics/stream-analytics-window-functions