Hi Brandon Another example, even if the idea is not to read but to modify the SGA... 4) To test particular conditions. E.g. in 10.2 I used to modify blocks in the buffer cache to check how/when checks performed by DB_BLOCK_CHECKSUM=FULL takes place, i.e. to produce "good" logical corruptions. Chris > -----Original Message----- > From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l- > bounce@xxxxxxxxxxxxx] On Behalf Of Anjo Kolk > Sent: Monday, August 21, 2006 8:03 PM > To: Allen, Brandon > Cc: tanel.poder.003@xxxxxxx; jeremiah@xxxxxxxxxxx; Oracle Discussion > List > Subject: Re: reading the SGA from my own program > > > There are a couple of resaons why this is (was) valid: > 1) Overhead > Sampling the v$SESSION_WAIT once a second is something that > can't be done with normal queries (not completely true) and mapping > different pointers. > A join of v$session_wait, v$session and v$sqlarea is rather > expensive if you want to find the current sql statement that causes > the wait, very cheap with > direct SGA attach. > 2) Information > Sometimes to get usefull info from V$ or X$ tables, one needs to > join them and even then it becomes not possible. By having access > to the SGA > one can join structures together that can't be done in normal > SQL. > 3) Geeky stuff > Really cool to do ;-) > > Anjo. -- //www.freelists.org/webpage/oracle-l