Details
-
New Feature
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
None
-
None
-
None
Description
To perform better joins against small fact tables (I believe tpch has a case like this) we want to add hash join as another join impl.
There are a variety of hash join strategies but I plan to start with a basic one where assuming one side fits entirely in memory.
Part of this is to look for a good off heap hash map impl or potentially write one if none seems fit.