| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> Re: 10g System statistics - single and multi
Christo,
You should separate things that are maybe happening on the disk level or disk cache level from the Oracle optimizer COST.
Oracle does not recognize is the read random disk I/O read or eventually sequential single disk read.
For Oracle all single block reads are measured equally if not proved differently.
What is happening after that at execution time neither Oracle or OS or I/O subsystem will give you measures, prediction or cost in front at the time COST is calculated.
Also, not sure how are you going to prove anything about the difference while ORacle does not know anything about that difference or you have statistics/events recognizing or counting the difference.
At the end, even Oracle developers are not building the neural network optimizer that will give you unpredictable execution plans because at the end you need to balance the gain of having simple deterministic (eve based on 100 of input parameters) useful algorithm and too complex non deterministic beast.
Regards,
Zoran Martic
-- http://www.freelists.org/webpage/oracle-lReceived on Fri May 20 2005 - 15:14:39 CDT
 |
 |