RE: Question - any parameter or hint to FORCE Physical Read (bypass buffer cache)

If there is caching on your disk farm AND there is a correlation such that the apparently "slow" node is more often causing the cache reads that then benefits the other nodes when they read, that node will appear slower. There is basically nothing you can do within the Oracle engine to make first reads on the disk farm faster.

Of course it could well be one of the many things mentioned in the thread regarding the Oracle engine. In the Morle papers mentioned, you also need to keep track of which cache layer he is talking about which you CAN pick up from context whenever he is not specific. Morle seems to always cut right to the heart of the matter and explain succinctly what is important.

mwf
-----Original Message-----

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Christopher.Taylor2_at_parallon.net Sent: Monday, April 22, 2013 3:58 PM
To: oracle-l_at_freelists.org
Subject: Question - any parameter or hint to FORCE Physical Read (bypass buffer cache)

I don't want to flush the buffer cache but need to verify disk IO speeds on 3 servers in my rac cluster. According to DBA_HIST_SQLSTAT, the elapsed times for some queries are significantly slower on one of the 3 nodes. I've done a cursory glance at slob and it seems a bit complicated for what I want to do here *if* there is an option to force physical reads and bypass the buffer cache.

Chris Taylor
Oracle DBA
Parallon IT&S

--

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

--

http://www.freelists.org/webpage/oracle-l Received on Tue Apr 23 2013 - 04:18:08 CEST