Sudden spike in physical reads for an update statement

From: vijay sehgal <vijaysehgal21_at_gmail.com>
Date: Thu, 6 Feb 2014 22:58:29 +0530
Message-ID: <CALQThVfH2U=1QY53U3kAxzHwmXOUsfiJLMjoNz9secFyiixYTw_at_mail.gmail.com>



Dear Experts,

I am trying to understand why an update statement having disk reads in the range of 90 per execution earlier went to 17.5K per execution.

The update statement is executed by Oracle as the records from Parent table are being deleted and child table has "on delete set null" rule.

I only have the execution plans now and they are same. The reports I have are generated by awrsqlrpt, the difference in buffers gets is not much but difference in physical reads is huge.

Where should I dig to get the cause of sudden spike in disk reads? I don't have access to the box, all I can do is request for reports (AWR / ASH). I won't be given 10046 of this query as it's from Live.

Any help to ask pointed question and get reports to find the cause are much appreciated.

Warm Regards,
Vijay Sehgal

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Feb 06 2014 - 18:28:29 CET

Original text of this message