Windowing in Kafka Streams and Flink SQL

Shakespeare once said that the ""Eyes are the window to your soul"". Had Shakespeare been a developer today, he may have changed that to ""Windows are the eyes into your data.""

Stream processing has become the de facto standard of working with data, with Kafka Streams and Flink being the top choices to implement an event streaming application. Responding quickly to any event is only possible when you can access those events as they happen. But in many cases, you're not concerned with one single event. Instead, it's a series of events within a given period that commands attention. In other words, it's essential to analyze events within discrete windows of time. Yet, with the different options available and the time semantics around them, windowing can be tricky to get right.

In this talk, I will cover the following topics for windowing in Kafka Streams and Flink SQL:

• Different window types (hopping, tumbling, sliding, etc.) and use-case application
• The semantics of time advancement, window closing, and the emitting of results
• Analyzing windowed results
• Testing strategies

Developers attending this presentation will gain an understanding of what windowing is in stream processing, the different types available to them, and some guidelines on when to apply which window type.