Influence of Nr of rows in rowset on Merge Join

Posted by & filed under Kettle.

Let's consider a transformation that merges two flows:

Here are some experiments with different Nr of rows in rowset:

So the speed of the Merge join depends on the parameter Nr of rows in rowset. It should be reasonably high (3000K).

Note that if the parameter is too high, the transformation might fail with exception: java.lang.OutOfMemoryError: Java heap space.