A particularly relevant side effect here is that a plan that uses a merge join rather than a hash join could have produced the results in a different order (even when joining the tables in the same order), and that could have a significant impact on the clustering_factors of the indexes created on the data, which could result in unexpected choice of indexes and surprising execution plans later on in the life of the application. Regards Jonathan Lewis ________________________________________ From: oracle-l-bounce@xxxxxxxxxxxxx [oracle-l-bounce@xxxxxxxxxxxxx] on behalf of Hans Forbrich [fuzzy.graybeard@xxxxxxxxx] Sent: 15 April 2013 02:55 To: oracle-l@xxxxxxxxxxxxx Subject: Re: Query does not work in RAC > In my case, the SQL should run only once, and I just changed the USE_HASH > to USE_MERGE and solved my problem ... Lucky. But do you know WHY? And whether this will impact anything in the future? Or will this lead to another hit-and-miss myth?-- //www.freelists.org/webpage/oracle-l