Microsoft Fabric Data Engineer Associate DP-700 Practice Question

The task requires a sliding (overlapping) window because results must be recomputed every two minutes for the previous ten-minute period. In Spark Structured Streaming a sliding window is created by passing both the window duration and the slide duration to the window function. Watermarking is added to admit late data up to a specified threshold. Therefore, the correct solution is to apply .withWatermark("eventTime","3 minutes") .groupBy(window(col("eventTime"), "10 minutes", "2 minutes"), col("deviceId")) which:

• Defines a 10-minute window that slides every two minutes, giving overlapping results that refresh as required.
• Adds a three-minute watermark so events that arrive within that lateness are still included.
  The distractors either create non-overlapping (tumbling) windows, omit the slide interval, set an incorrect watermark, or fail to add any watermark, thus violating latency or correctness requirements.