Windows in stream processing | Course in Spark Structured Streaming 3.0 | Lesson 6

[NuhilMehdy-mzgi]

Hi Amit,
In your 'Sliding (Hopping) Windows' slide, at 12:05 timestamp you have a window of length 12:00-12:05 which makes sense since the first trigger happened at 12:05 but there is nothing earlier than 12:00 to make a 10 minutes long window. However, in your 'Examining Slidding Window' slide, at 12:05 we have a trigger but the window values are shown as 12:00 - 12:10. My question is - at 12:05, should not the window be 12:00-12:05 or at least marked as 11:55-12:05? Or does it mean that at 12:05 the window is stretched on both sides of 12:05 and it assumes a window of 12:00-12:10. Could you please explain this a bit more to understand the concept more consistently?
Thanks for your video series :)

[baskarshanmugam]

Amit, Thanks for the detailed video. You mentioned about custom-size window and I like to know do you have any sample implementation to show how it is done.

[dharmendersingh]

Hey Amit,
Aggregation would be done at batch interval right and not the sliding window interval ?

[NdamuleloNemakh]

For custom sized windows, what happens if you do not receive an exit event?

[esakkisundar]

How to handle duplicate events in sliding window.