Jonathan also mentioned to me that the cost estimate should be = multiplied by 'sreadtim' when running 9i with system statistics (cpu costing) enabled. The cost = multiplied by this value is the assumed time to complete. Cost is not the time estimate itself and this is what I was trying to = discern from Jonathan's earlier post. What I understand Jonathan was trying to say = was that cost is the value that should be multiplied by a time value estimate to = determine the estimated time to complete. The issue I still find for myself is that I rarely see this estimated = value be reliably close to the actual value for response time. Perhaps in the perfect = world, and I guess that's where the optimizer thinks it lives :), it would. But, even on a = test database where I am the only user executing a single query, I don't often see the = costed time estimate match the actual. I just wish the optimizer was perfect....then again, if that were the = case, many of us would have to find other ways to fill our time currently allotted to = query optimization. :) Karen Morton Hotsos Enterprises, Ltd. http://www.hotsos.com Upcoming events at http://www.hotsos.com/education/schedule.html =20 -----Original Message----- From: Mladen Gogala [mailto:gogala@xxxxxxxxxxxxx]=20 Sent: Friday, January 14, 2005 3:23 PM To: karen.morton@xxxxxxxxxx Cc: oracle-l@xxxxxxxxxxxxx Subject: Re: Which plan is better - what COST really means ? On 01/14/2005 05:43:53 PM, Karen Morton wrote: > In what time measurement is the cost? Seconds, centiseconds, =3D=20 > microseconds? =3D20 The measurements are in centiseconds, that is soft clock ticks. That is = one of the few things that wasn't converted to microseconds with =20 oracle9i --=20 Mladen Gogala Oracle DBA -- //www.freelists.org/webpage/oracle-l