The parallel data processing system Apache Flink features APIs for different use cases. There is the DataSet API for batch-style programs, the DataStream API for real-time programs, and the Table API/SQL for analytical use cases on both batch and streaming data.

For some time now, the Flink community has discussed how we can reduce the number of APIs by unifying them. In this talk I will discuss the motivations for wanting more unified APIs, what it actually means to unify APIs and how we plan …