As an Addendum to my posting Optimizer Stats: Treat some tables different than AutoTask does , I’d like to refer to this interesting and helpful article from the Oracle Optimizer Dev Group.
This article describes (for 10g & 11g) how to remove existing histograms and exclude tables from getting histograms in the future, when the Automatic Optimizer Stats Gathering takes place. A particular case why that is necessary would be if a skewed varchar2 column has more than 32 characters – but the first 32 characters are identical. Then the histogram will strongly mislead the Optimizer regarding the cardinality of that column.
Pickleball spielen 