Sliding windows define all possible intervals of equal duration and output aggregations based on actual event stream activity. Sliding windows can be helpful when an event stream has unpredictable or spiky volume activity. They can help reduce the output during periods of inactivity or when the aggregation results remain unchanged.
The following example shows the average speed and count of cars per city. The aggregates run over every 10-second window and return records every time a computation result changes. When a new car appears, a new record with 10-second computations is produced, with the car count incremented by one. ...
