Logical Standby: Solving the Uniqueness Problem
From: fmh <fmhabash_at_gmail.com>
Date: Fri, 26 Apr 2013 10:50:34 -0400
Message-ID: <CADpeV5xUD_EufABvejLGRNRbopogoFE4m_mHkVvLpn7T5OzAHg_at_mail.gmail.com>
As part of creating a logical standby, there is a step where tables without unique identifiers (PK/UK) columns must have a way to uniquely identify a row. I understand it is not mandatory but at a minimum, tables with bad_column=Y MUST have PK/UK constraint. If not, I risk having duplicate rows in the logical standby.
Creating a PK/UK on a table, even though technically a trivial task, it is not from an application standpoint. First, the application is not aware of it. As such, transactions will fail if it tries to insert/update row that violates the constraint. Since it was not part of the original app design it will not populate it.
Thank you.
Date: Fri, 26 Apr 2013 10:50:34 -0400
Message-ID: <CADpeV5xUD_EufABvejLGRNRbopogoFE4m_mHkVvLpn7T5OzAHg_at_mail.gmail.com>
As part of creating a logical standby, there is a step where tables without unique identifiers (PK/UK) columns must have a way to uniquely identify a row. I understand it is not mandatory but at a minimum, tables with bad_column=Y MUST have PK/UK constraint. If not, I risk having duplicate rows in the logical standby.
Creating a PK/UK on a table, even though technically a trivial task, it is not from an application standpoint. First, the application is not aware of it. As such, transactions will fail if it tries to insert/update row that violates the constraint. Since it was not part of the original app design it will not populate it.
If you have worked with logical standby's, how did you resolve this issue?
Thank you.
--
http://www.freelists.org/webpage/oracle-l
Received on Fri Apr 26 2013 - 16:50:34 CEST