Re: how to force an index range scan?

When creating histograms, you do need to know a bit about the data to decide on the bucket count. There is a good reason for Oracle's default bucket count being 75 - anything much lower, and you generating statistics with little value. Unless 99% of the data fall into 19 of your 20 buckets, the buckets add little value to the column level stats.

If you post bucket number and end-count for the histogram along with the total row-count, min, max, distinct value and number of nulls of the column and the value you were after, I'll see if I can justify Oracle's 'error' for you.

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

gdas_at_my-deja.com wrote in message <8qgecl$nrt$1_at_nnrp1.deja.com>...


>Sorry for so many posts here.  But I seem to have found the problem.

>Don't quite understand it, but it will do for now.

>

>I was analyzing the f_misc table sampling 20 percent and creating a 20

>bucket histogram on all indexed columns.  When I removed the histogram,

>I got my index range scan.  I tried it with different predicate values

>for account_id and they all come back fast.  Very strange.  I knew

>histograms don't always help, I guess they can in fact hurt you

>sometimes as well.

>

>Thanks for your time.

>Gavin

>

>

>

>Sent via Deja.com http://www.deja.com/

>Before you buy.