Details
-
Sub-task
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
I think there are two pieces:
- the separate parallel unit (always a Scan for Phoenix)
- the reader class to instantiate that will interpret the results.
It be pretty easy to write code that combines multiple Scans together too, based on the start row/end row in the case that Phoenix over-parallelizes due to guideposts.
FWIW, even the Phoenix DML (UPSERT SELECT and DELETE) fit this model of a Scan plus the reader code which would push the data back out for the writes (see UpsertCompiler and DeleteCompiler).