Stream processing is how IoT backends react in near real time. A framework such as Apache Flink, Kafka Streams or a cloud equivalent reads events from a message queue, keeps state per device or per window, and emits results continuously: a rolling average, a threshold crossing, a count of devices that went silent in the last few minutes. It generalizes the simple conditions of a rule engine to logic with state and joins.
The trade-off against batch processing is complexity. Streaming systems must handle late and out-of-order events, exactly-once semantics and state recovery after failure, all of which take engineering effort. Many pipelines use streaming for alerts and batch for reports.
Part of the IoT Glossary. Spotted an error or a missing term? Tell us.