When running the statement in relations.sql, our Redshift cluster query planner will, seemingly at random, expect to use a nested loop join. This reflects in extended execution times, adding anywhere ...