Data Streaming

Data streaming means processing events continuously the moment they arrive, rather than collecting them and processing in batches at intervals — the basis of any “live” view.

Streaming stands against batch processing, and the choice between them is a trade-off, not an absolute superiority. Streaming raises data freshness and makes a real-time view possible, but is more complex and must cope with data arriving out of order or late. Batch processing is simpler and sufficient for a historical report but is inherently delayed. What a user should grasp is that this very difference is why tools have both a “realtime” and a “historical” view: the live view comes from the stream and may be slightly corrected later, the historical view from settled processing. So the live number and the historical number are not meant to be exactly equal, and that small inconsistency is a sign of the system's health, not its error.