Windowing is an important query language extension that Stream Analytics provides to group events by time intervals, enabling aggregations of data streams. Other language extensions help with additional temporal aspects of stream data processing, including the source of the timestamp on which windows will be calculated, and the settings governing potential timestamp conflicts.
In a streaming system, a timestamp is the most fundamental data element in an event, and thus every event must have one in order to be processed or queried. In a simple streaming system, we can guarantee this by defining the moment each event arrives in the event stream as its identifying timestamp. (For Stream Analytics, the event stream is either Event Hub or IoT Hub.) Arrival Time is the default identifying timestamp of events. However, Stream Analytics provides a mechanism to choose a timestamp, known as Application Time, based on a column in the payload instead. Furthermore...