Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Migrating 8i OPS to RAC 9.2

RE: Migrating 8i OPS to RAC 9.2

From: Jamadagni, Rajendra <>
Date: Mon, 23 Jun 2003 08:50:45 -0700
Message-ID: <>


do you think there is no downside to bring down this parameter? I know we had a peculiar process that really zoomed when we had set _fairness_* parameter to 1, but didn't go ahead with it in production because we couldn't get a satisfactory answer of its downside.


Rajendra dot Jamadagni at nospamespn dot com All Views expressed in this email are strictly personal. QOTD: Any clod can have facts, having an opinion is an art !

-----Original Message-----
From: K Gopalakrishnan [mailto:[EMAIL PROTECTED]] Sent: Monday, June 23, 2003 9:49 AM
To: Multiple recipients of list ORACLE-L Subject: Re: Migrating 8i OPS to RAC 9.2


I am not aware any of the documentations which explains this  cache fusion to ping pong feature(!).  There are 3 components in the typical CR prosessing..

Let us assume there is a resource R which is mastered by the instance A and owned by the instance B on X mode. And also we assume the resource R is requested by Instance C. In this case the requester enquires the status of that resource to the master database and got to know that is owned by instance B.  So now it is Instance B's respoisibility to constuct the CR and send it to Instnace C.

Here, there is something called light work rule (X$KCLCRST.LIGHT?) which decides the CR construction and block transfer over interconnect
(again the CR processing for S to N locks are different based on the
setting of _cr_grant_local parameter) or thru the disk transfer.  
Basically the current holder of the resource maintains a fairness counter, which is incremented every time it sends  CR copy over the interconnect to the requester and there is a threshold  for the number of CR copies created for that resource. Once the ceiling is hit, instead of creating the CR copies, the LMD simply downconverts the lock to NULL and informs the convertion to Intance A.

In simple terms it is like a normal OPS ping. The owner downgrades the lock and the requester reads from the disk after getting approval from the lock master. The threshold is controlled by the parameter _fairness_threshold and IIRC that defaults to 4 or 5. So every 6th (or 5th) CR request will most of the times results in a PING and I have seen good performance improvements in most of the RAC databases by changing this parameter.

Best Regards,
K Gopalakrishnan

(Currently I am in a country (for a week) where I have very limited
access to the internet , So I may not be able to reply if you have any more questions And I don't have any oracle database/documentation to test/verify. So please take the advise with a pinch of salt !)

Please see the official ORACLE-L FAQ:
Author: K Gopalakrishnan

Fat City Network Services    -- 858-538-5051
San Diego, California        -- Mailing list and web hosting services
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L

(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing). *********************************************************************This e-mail message is confidential, intended only for the named recipient(s) above and may contain information that is privileged, attorney work product or exempt from disclosure under applicable law. If you have received this message in error, or are not the named recipient(s), please immediately notify corporate MIS at (860) 766-2000 and delete this e-mail message from your computer, Thank you.*********************************************************************1
Received on Mon Jun 23 2003 - 10:50:45 CDT

Original text of this message