Streaming stock prices from Pub/Sub into Dataflow require computing a moving average of the past 30 seconds every 5 seconds. How should you configure windows and triggers?

The problem requires a moving average over the past 30 seconds, updated every 5 seconds. This is a classic use case for a sliding window. A sliding window of 30 seconds duration and 5 seconds period ensures that the window slides forward by 5 seconds each time, processing the last 30 seconds of data. Emitting results AfterWatermark.pastEndOfWindow() is appropriate for producing a result once the system believes all data for that window has arrived, which is standard for batch-like processing within a stream. Incorrect options: Fixed windows do not provide a moving average; they process disjoint segments of data. The other sliding window option has an incorrect duration (5 seconds instead of 30 seconds) and an inappropriate trigger for this requirement.