>> I am wondering, why the historical data (real result of the query) does not affect the execution plan. For example, If I ran the query 10 times I always get around 6500 instead of 5.5 million in the top most execution plan node. The likely accurate, if cheeky, answer is simply that no one has taken the time to implement what at face value would seem to be an extremely complex behavior. As to the rest you should probably gather up some additional details and post this over on performance. There is likely a middle ground between your idea and the status quo but that ground cannot be evaluated unless examples of possibly misbehaving queries are provided. David J. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
