Flink gives us tools to slice, transform, aggregate, recombine, and otherwise process streams.
Note
Stream processing is actually hard and expensive to get right. Many problems that seem like stream processing problems can actually be reduced to batch processing problems where you'd use something like Spark or (if you're ancient enough) Hadoop.
Before embarking on a stream processing solution, ask yourself the critical question: "do I really need real-time latencies?". For many problems, the answer is no and the engineers after you will thank you for saving them the ops headache.