Out of order and late-arriving events

The Out of order policy defines a time allowance for events arriving at the Stream Analytics job out of order. It governs how long the Stream Analytics job will buffer events and, within that grace period, corrects the order according to the application timestamp set in the query. While the Out of order events policy is a helpful mechanism by which to manage timing conflicts, it does introduce latency equal to the time allowance duration itself.

Due to connectivity and networking reasons, events generated by source application will arrive out of order. For example in IoT scenario, the connected devices can suffer from intermediate connectivity in which case a set of data will be held on the device that is waiting for a connection to be re-established before the burst is transmitted. Forecasting intermittent device connectivity reliably is a lot more difficult. To address these scenarios, Azure Stream Analytics provides the ability to define a threshold...