Which window type is dynamic, non-overlapping, and data-driven for aggregating key-based events into a session?

Which window type is dynamic, non-overlapping, and data-driven for aggregating key-based events into a session?

• A. Tumbling Window
• B. Hopping Window
• C. Sliding Window
• D. Session Window

Answer: (D)

Session windows are dynamic, non-overlapping, and data-driven, which is exactly what you need for aggregating key-based events into a session. They start when a new event for a key arrives and end after a period of inactivity, defined by a session gap. Because the end of a session depends on the observed activity (data-driven), the length can vary from one key to another and even from one session to the next. They are non-overlapping for a given key because once a session closes, any new activity starts a fresh session rather than extending the previous one. This makes them particularly well-suited for modeling user or key-based behavior: you accumulate events into a session as long as activity continues, and you emit a summary when a lull happens. Other window types are fixed or overlap: tumbling windows have fixed, non-overlapping time spans; hopping windows advance with fixed steps and can overlap; sliding windows overlap continuously. Hence, session windows best satisfy the criteria of being dynamic, non-overlapping, and data-driven.

Session windows are dynamic, non-overlapping, and data-driven, which is exactly what you need for aggregating key-based events into a session. They start when a new event for a key arrives and end after a period of inactivity, defined by a session gap. Because the end of a session depends on the observed activity (data-driven), the length can vary from one key to another and even from one session to the next. They are non-overlapping for a given key because once a session closes, any new activity starts a fresh session rather than extending the previous one.

This makes them particularly well-suited for modeling user or key-based behavior: you accumulate events into a session as long as activity continues, and you emit a summary when a lull happens. Other window types are fixed or overlap: tumbling windows have fixed, non-overlapping time spans; hopping windows advance with fixed steps and can overlap; sliding windows overlap continuously. Hence, session windows best satisfy the criteria of being dynamic, non-overlapping, and data-driven.