Re: Statspack Interpretation

From: <mccmx_at_hotmail.com>
Date: 20 Apr 2006 06:59:28 -0700
Message-ID: <1145541568.722347.258210@v46g2000cwv.googlegroups.com>

Yeah - i think you are on the right lines...

but I think the main reason is because the statements that do the most I/O are re-reading the same disk blocks and they benefit from the very large SAN cache we have. Conversely, those statements which hit relatively fewer physical blocks which are running infrequently are taking longer because thoses blocks arent in the SAN cache.

this theory is backed up by my statspack report....

I am seeing some DSS queries which are run once a day which aren't visiting that many physical blocks but they account for a large proportion of the I/O wait times.

It all makes sense now..

Cheers for your help.

Matt Received on Thu Apr 20 2006 - 08:59:28 CDT