Re: 10g CBO and how to determine cardinality on INDEX_FFS?

So the cost makes sense..

(LEAF_BLKS/MBRC) * (MREADTIM/SREADTIM)
 +
#CPUCycles / (cpuspeed * sreadtim_in_microseconds)

• I obtained #CPUCycles from the plan_table.cpu_cost by doing the explain plan cmd... as you mentioned in your book.

IOCOST = (39550/125) * (26.806/4.245) = 1997.97842 +
CPUCOST= 281652952 / (198 * (4.245*1000)) = 335.097681 = 2333.0761 which is very close to 2335 as calculated by the optimizer.

I'm still having some difficulties calculating the cardinality though. This is probably since I have a histogram on the product_id column. I've checkout
chapter 7 on histograms, but can't make the numbers add up.

This is what's currently in my user_tab_histograms view. There are only 2 values for product_id in that table.

ENDPOINT_NUMBER|ENDPOINT_VALUE
---------------|--------------

 1354519|     430657811
 1943369|     438075481

if you have a moment, any additional hints would be greatly appreciated.
I think I'm still not sure of what are the standard rules for calculating the
cardinality on columns with frequency histograms.

thanks again
--peter Received on Sat Nov 19 2005 - 00:52:46 CST