The tables are setup so that for 1 join condition all rows match,
but with 2 join conditions 1/5 of the rows match. Since we have a
limit, for the 2 join conditions more rows (all rows of t2) need
to be processed to fulfil the limit of 2000.
The spec is not added to the tracks so it's not executed periodically but could be used
in the future for such benchmark comparisons.
The tables are setup so that for 1 join condition all rows match, but with 2 join conditions 1/5 of the rows match. Since we have a limit, for the 2 join conditions more rows (all rows of t2) need to be processed to fulfil the limit of 2000.
The spec is not added to the tracks so it's not executed periodically but could be used in the future for such benchmark comparisons.