> if it is acceptible then why touch it? Doesnt matter if you do > 10000000000 LIO's per query if it works for you and the users are > happy ..Unless you could provide identical performance at 20% of the cost. Cary Millsap Hotsos Enterprises, Ltd. http://www.hotsos.com * Nullius in verba * Visit www.hotsos.com for curriculum and schedule details... -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of David Sharples Sent: Wednesday, April 20, 2005 1:12 PM To: Oracle-L Subject: Re: Good Performance The only real, metric you should be looking at is 'End user response time' if it is acceptible then why touch it? Doesnt matter if you do 10000000000 LIO's per query if it works for you and the users are happy You then have the task of making sure it is scalable. Thats why you need a test bed which you can simulate users on. If you test it and find it doesnt scale, find the problem processes and fix them before it goes life and you add another 1000 users on there On 4/20/05, Wiegand, Jeff <Jeff.Wiegand@xxxxxxxxxxx> wrote: > Could someone give me some "benchmarks" defining good performance? >=20 >>=20 > What is a good benchmark for determining if a query is doing too many > buffer reads, or too many disk reads? Is 10,000 buffer reads considered > high, or is it 1,000,000? Or does it matter if, on the overall, things > kick butt. -- //www.freelists.org/webpage/oracle-l -- //www.freelists.org/webpage/oracle-l