An image conveys much more than a paragraph and because of that reason we will start this section with a figure. The functioning of Flink is as shown in the following figure:

Flink is capable of taking in both batch and stream data. It operates on batch data as if it is another form of stream data and this itself is quite a unique feature of Flink. We have in one of the chapters in Part 1 explained a bit on Kappa Architecture was explained, in which all data is being considered and dealt with stream data and Flink uses that exact principle in its architecture and implementation.

In the preceding figure, both types of data (batch and stream) from various source systems gets into Flink. The Flink program submits the job and using master and worker, deals with these data and produces output.