Re: UPDATE statement - cpu and elapsed time almost the same

From: Andy Sayer <>
Date: Sun, 19 Nov 2017 10:47:46 +0000
Message-ID: <>

Hi Amir,

The tkprof summary you’ve shared shows that you didn't do physical IO, this is corroborated by the lack of reads column in your row source execution stats plan (assuming you didn't intentionally leave it off).

The number of buffers used here just doesn’t add up to the time you are experiencing (it should really be less than a second). Something else must be going on.

Is the problem repeatable? Doing a snapshot of v$sesstat before and after the statement, then taking the differences could reveal some clues.

Are there any further statements in the tkprof report? Perhaps a trigger is firing? Perhaps a foreign key is being checked? It could be worth sharing the raw 10046 trace (uploading it to a text file sharing site rather than attaching it here, to save our inboxes)


Received on Sun Nov 19 2017 - 11:47:46 CET

Original text of this message