Hands-On Reactive Programming in Spring 5

Backpressure handling

Even though the Reactive Stream specification requires backpressure to be built into the communication between the producer and consumer, it is still possible to overflow the consumer. Some consumers may innocently request an unbound demand and then fail to handle the generated load. Some consumers may have hard limitations for the rate of incoming messages. For example, a database client may not insert more than 1,000 records per second. In that case, events batching techniques may help. We covered that approach in the Batching stream elements section. Alternatively, we can configure a stream to handle backpressure situations in the following ways:

• The onBackPressureBuffer operator requests an unbounded demand and ...