Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: x$kghlu, Steve's shared_pool_lru_stats

Re: x$kghlu, Steve's shared_pool_lru_stats

From: Boris Dali <boris_dali_at_yahoo.ca>
Date: Mon, 8 Nov 2004 11:41:07 -0500 (EST)
Message-ID: <20041108164107.71496.qmail@web41412.mail.yahoo.com>


Hi Tanel,

Thanks for the reply. Yes, the moment shared_pool_size is set to 250MB or higher, Oracle responds by setting _kghdsidx_count to 2 (meaning shared pool is divided into 2 heaps/protected by 2 child latches/has 2 LRU lists?).

Incidently, it looks like the decision to split the shared pool is made based on the shared_pool_size parameter alone, not the derived value of the shared pool, i.e. having shared_pool_size=200MB and processes=8000, which yeilds "variable+fixed"=593MB (much higher than say shared_pool_size=250MB, processes=150) doesn't qualify for splitting.

I toyed around with the size of the shared pool in the attempt to get 3 heaps, but to no avail. Even setting shared_pool_size to 20GB didn't produce 3 heaps. Just curious how does one get 7?

Also a more practical question (I think). If my "optimal" shared pool size is around 250MB (i.e. at 150-200MB I get ORA-4031), would I be better off with 240MB and a single heap or at 250MB and dual heaps?

Thanks,
Boris Dali.


Post your free ad now! http://personals.yahoo.ca
--
http://www.freelists.org/webpage/oracle-l
Received on Mon Nov 08 2004 - 10:36:57 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US