RE: IO testing - CALIBRATE_IO (11g) vs. other (10g)
From: Michael McMullen <ganstadba_at_hotmail.com>
Date: Thu, 4 Apr 2013 11:28:31 -0400
Message-ID: <SNT130-W538217B0B90CC8FDB1FDEBA6D90_at_phx.gbl>
create your own small db on prod for slob. there are specific parameters you need to ensure that it's performing physical i/o. slob requires many schemas though I think someone has rewritten some of it to only use one. you don't want to mess with your prod. do you have a dev server that points to the same san as prod with the same plumbing? others might chime in but I don't think you need to worry about #'s of cpu, that all gets set to 2 with slob. I would definitely run slob on the new platform to get a handle on how it works and to get your numbers.
Date: Thu, 4 Apr 2013 11:28:31 -0400
Message-ID: <SNT130-W538217B0B90CC8FDB1FDEBA6D90_at_phx.gbl>
create your own small db on prod for slob. there are specific parameters you need to ensure that it's performing physical i/o. slob requires many schemas though I think someone has rewritten some of it to only use one. you don't want to mess with your prod. do you have a dev server that points to the same san as prod with the same plumbing? others might chime in but I don't think you need to worry about #'s of cpu, that all gets set to 2 with slob. I would definitely run slob on the new platform to get a handle on how it works and to get your numbers.
-- http://www.freelists.org/webpage/oracle-lReceived on Thu Apr 04 2013 - 17:28:31 CEST