More about Remote AE

The term flowability refers to the principle that all dataslices in a query must be actively and simultaneously processed. Failure to satisfy this requirement risks poor performance at best, deadlock at worst. However, this does not mean that there cannot be a delay, even a long one, between input and output requests. Records should not be processed from one dataslice while ignoring others; the dataslices must be processed in parallel.

The result is that every dataslice of every query requires its own thread of execution, using either a thread or a process fork. Anything less—for example, a task switching algorithm—is not sufficient since one dataslice could be in a blocked state while data is available on another.

For languages that support threads, the use of a thread pool can improve performance.