KIP-450 — Sliding Window Aggregations in the DSL

Discussion Kafka 2.7 Streams

Implements sliding windows in Kafka Streams where each record creates exactly two new windows (one ending at the record's timestamp, one starting at it), enabling precise window boundary semantics without size-aligned bucketing. Tumbling and hopping windows align boundaries to fixed grid intervals, which loses the fine-grained event-time resolution needed for use cases like computing metrics over a sliding N-millisecond window relative to each event.

Details