Apache Flink is designed to process data in real time, creating robust stream processing pipelines through understanding the what, where, when, and how of data.
The ‘what’ of data processing includes determining the computations to perform, whether they are simple aggregations or complex transformations like building machine learning models.
Understanding ‘when’ refers to utilizing triggers and watermarks to determine when results can be emitted and how data is finally triggered within the stream.
The ‘how’ aspect encompasses the types of accumulation methods used in processing, such as discarding, accumulating, and retracting, impacting how results are computed over time.
Collection
[
|
...
]