Re: Simple Insert tuning

Thanks Norman and Wilson for guiding the process Sanjay

    On Thursday, November 30, 2017 5:51 PM, Norman Dunbar <oracle_at_dunbar-it.co.uk> wrote:  

 Evening Sanjay,

I recently had a similar problem to work out for a friend who is a developer but doesn't/didn't have a dba to consult. His problem was an insert taking 2048 (spooky number) seconds to insert the first row, all the following rows were almost instant.

A 10046, level 12 trace showed the problem was related to dynamic stats being gathered prior to the first EXEC of the insert. The stats were being gathered on various SYS objects. It appeared that the sys schema and dictionary stats were never gathered on his database. After gathering (both) the insert problem vanished.

I fully agree with the other poster (sorry, forgot your name) who suggested tracing as the way to find out. It will show you exactly where Oracle spent its time.

Good luck.

Cheers,
Norm.
--

Sent from my Android device with K-9 Mail. Please excuse my brevity.    

--

http://www.freelists.org/webpage/oracle-l Received on Fri Dec 01 2017 - 00:36:58 CET