Thanks for the links, but I am still not convinced as to why oracle behaves like this, according to my understanding its thinking somethink like


Avg No Of Non popular values per bucket=16.92425926283973353

Number of values per bucket=42

So Avg Number of popular values per bucket ~~ 23

It even isnt recognizing when queried with the very low (repeating values) & its going for FTS. I think 145 bucket size isnt sufficient for distinct count of 2000+ to abloutely determine the skewness of the column.

can somebody correct if i am wrong, or even suggest.

Well i am now getting convienced that Histograms on highly/averagely spread values should not be taken histograms. Well i only wanted to see how Oracle will calculate Evenness of the column.

Histograms will well be useful if distinct count of the value is well less than 254.

I would like to get suggestion anyhow ( from probabaly slimdave, pando, jurij, dapi, tamil, chris )

Abhay.