System Statistics oracle9i

Hi  

How do you handle system stats when your gather_system_stats returns MREADTIM < SREADTIM?

Do you live with the optimizer using the adjusted dfmrc value?

Apply some calculation to SREADTIM to generate a larger MREADTIM?

Ignore all stats where MREADTIM<SREADTIM and pick some average for the remainder?

Or what?

We have dynamic memory buffer caching turned on here (HP-UX ia64 itanium). Is this the likely reason why often MREADTIM<SREADTIM? Or do I blame the SAN?  

... and ...  

My gathered system statistics are much more variable that I would have thought. For the same time period that is.

Is it possible that the two instances on the same server will affect each other's system statistics?

Is there likely to be a problem with gathering system stats for both instances at the same time?

Even so, what do you suggest to handle this variation? Plan A is to take the average of all of the stats.  

Instance1: 08:30-14:30  

STATID SRDTM MRDTM CPU MBRC MXTHRD

• ---------- ---------- ---------- ---------- ----------

D27JUL0830 .995 1.13 1194 6 94699520 D26JUL0830 1.378 1.649 1190 6 34324480 D25JUL0830 1.482 1.088 1191 5 386048 D22JUL0830 1.67 .723 1185 5 13148160   Instance2: 08:30-14:30  

STATID SRDTM MRDTM CPU MBRC MXTHRD

• ---------- ---------- ---------- ---------- ----------

D27JUL0830 4.065 10.643 1120 30 68080640 D26JUL0830 .659 7.064 1194 31 38184960 D25JUL0830 .622 9.988 1198 28 23548928 D22JUL0830 1.558 9.757 1195 29 1528832   Thanks

Trevor      

DISCLAIMER:
Disclaimer. This e-mail is private and confidential. If you are not the intended recipient, please advise us by return e-mail immediately, and delete the e-mail and any attachments without using or disclosing the contents in any way. The views expressed in this e-mail are those of the author, and do not represent those of this company unless this is clearly indicated. You should scan this e-mail and any attachments for viruses. This company accepts no liability for any direct or indirect damage or loss resulting from the use of any attachments to this e-mail.

--
http://www.freelists.org/webpage/oracle-l