RE: shared pool size / library cache locks / too big ?
Date: Sun, 2 Jun 2013 14:04:00 -0500
Message-ID: <F05D8DF1FB25F44085DB74CB916678E887A03A5F75_at_NADCWPMSGCMS10.hca.corpad.net>
AMM is on but we have a floor for both shared pool and buffer cache. Buffer Cache is approximately 70gb - each server in the cluster has 128GB ram.
Checking DBA_HIST_ACTIVE_SESS_HISTORY for a 24-hour period on node2 there are 15K DISTINCT SQL_IDs.
There are 4K distinct FORCE_MATCHING_SIGNATURES over the same period (so LOTS of similar sqls with different literals).
Chris
From: John Hurley [mailto:hurleyjohnb_at_yahoo.com]
Sent: Sunday, June 02, 2013 1:40 PM
To: Taylor Christopher - Nashville
Subject: Re: shared pool size / library cache locks / too big ?
Are you using the automated memory management stuff and letting the shared pool grow to 22 gb ( not a big fan at all of the automatic stuff ) or is it hard coded to be that large? Most people probably "do not need" / "do not use (well)" a shared pool more than 4 or 5 gb. How many distinct sql statements are in your shared pool? How many are the same except for literal values?
How large is your buffer cache?
How large is the servers physical memory?
In general for many of the questions that you submit here supplying more background information rather than just a couple of small tidbits might contribute to getting a better discussion.
-- http://www.freelists.org/webpage/oracle-lReceived on Sun Jun 02 2013 - 21:04:00 CEST