...

Currently, Flink provides the AsyncIAsync I/O API to facilitate users to handle time-consuming I/O operations at lower cost and asynchronously to improve the throughput and response time of the program, but limited on retry support: the user interface AsyncFunction only provides an one-time callback handle, there is no good way to fire a retry based on asynchronous results(e.g. empty results or errors that asyncClient cannot handle, etc.).

Consider such an use case: user program does access external data via asyncIAsync I/O, some data may not be updated in external systems in a timely manner(offen happens due to the lack of a global coordinator),  and the user want do delayed retry when encounter such unexpected 'missed' data, but can't elegantly implement through the current AsyncFunction (asyncInvoke and callback functions are executed synchronously by the main thread, which is not suitable adding long time blocking operations, and introducing additional thread will bring extra complexity for users).

...

Public Interfaces

Use Case Via Current AsyncIAsync I/O API Without Retry

An example HBase AsyncFunction:

...