Haven't looked at STATSPACK in years, but it used to classify some events as "idle" (their names were stored in a table named STATS$IDLE_EVENTS, I think), and it wouldn't gather information about them.
I'll bet the other 8 mins is in one of those untracked "idle" events like "SQL*Net message from client".
On 2/17/15 17:23, Ram Raman wrote:
ListBelow is a copy and paste of a section of statspack from a DB on a v11.2 on windows. The DB Time of more than 17 mins compared to DB CPU of 2.4 mins got me scratching my head. I feel that it means there is lots of wait somewhere. But, the top most wait of dbf seq read is only about 7 mins. Where did the other 8 mins go?******************************************************************************* Snapshot Snap Id Snap Time Sessions Curs/Sess Comment~~~~~~~~ ---------- ------------------ -------- --------- ------------------Begin Snap: 4132 16-Feb-15 22:00:27 54 9.1 End Snap: 4143 16-Feb-15 23:00:27 51 10.1 Elapsed: 60.00 (mins) Av Act Sess: 0.3 DB time: 17.38 (mins) DB CPU: 2.44 (mins) Cache Sizes Begin End ~~~~~~~~~~~ ---------- ---------- Buffer Cache: 3,696M Std Block Size: 8K Shared Pool: 224M Log Buffer: 18,240K
-- //www.freelists.org/webpage/oracle-l
