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

Home -> Community -> Usenet -> c.d.o.server -> IO issues with Oracle Parallel Server

IO issues with Oracle Parallel Server

From: <maozmind_at_my-deja.com>
Date: Sun, 28 Nov 1999 10:42:05 GMT
Message-ID: <81r0tt$hmu$1@nnrp1.deja.com>


We installed OPS (Sun Solaris, Oracle 8.0.6) to work with our application. We are facing performance issues compare to our 1 node settings. The main thing we found is a high rate of IOs, both on data and index tablespaces. We tried to tune the database parameters
(GC_FILES_TO_LOCKS and other parallel specific parameters), but haven't noticed any change. The only change we noticed was after we allocate extent to one of the 'bottle neck' tables, on separate
datafiles and instances as follows:
ALTER TABLE activecalls ALLOCATE EXTENT (SIZE 100M DATAFILE 'dat25' INSTANCE 1);
ALTER TABLE activecalls ALLOCATE EXTENT (SIZE 100M DATAFILE 'dat28' INSTANCE 2);

This change reduced contention, and the IO rate decreased dramatically, performance increased a little.
This table has transactions on both nodes - INSERT/UPDATE/DELETE transactions on high rate. We tried to do the same on the table indexes, i.e. separate on two datafiles, one per each instance.
but in this change there was no improvement on the IOs made on this tablespace.

Any ideas why this trick worked with the data datafile, but not with the indexes? Any hits for such
a problem ?

Thanks,
Maoz

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Sun Nov 28 1999 - 04:42:05 CST

Original text of this message

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