Using DBMS_STATS to estimate statistics on a table . I created a table with
50,000 rows and it has one index . That column
has 5 distinct values . I ran a command to estimate statistics on that table
with skewonly option . It still went ahead and
created histograms . I was under the impression that since data distribution
is not skewed it will not create histograms .
Secondly , how does it know that data distribution is skewed or not skewed
using column stats generated .
I also want an opinion on how to decide whether to gather column stats for
all columns or just for indexed columns .
How to decide ?
Received on Wed Aug 25 2004 - 21:13:44 CDT
