Cary, Thank you a lot. I could not cover all test cases yesterday so easy because of no time and non-easy ways to trace parallel slave processes. So I asked you/oracle-l list. I just wanted to get more info about the direct path read associating with the parallel slaves from the master. What is on the net is very scattered and in some cases wrong. Also, very hard to find out anything about it except usual copied sentences from Oracle docs. But it looks my old knowledge about it did not change. So all blocks are going always to the PGA regardless of the cache. Regards, Zoran --- Cary Millsap <Cary.Millsap@xxxxxxxxxx> wrote: > I suspected that saying "never" might get me into > trouble. :) > > I think that 2 is false for direct path read, > because I believe that direct > path reads are not buffer cache managed. I should > have said, more precisely, > "Oracle will never PIO a block /into the buffer > cache/ that's already in the > buffer cache." > > > Cary Millsap > Hotsos Enterprises, Ltd. > http://www.hotsos.com > * Nullius in verba * __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new Resources site http://smallbusiness.yahoo.com/resources/ -- //www.freelists.org/webpage/oracle-l