Re: Recovery Manager (RMAN) problem

From: Robert Klemme <shortcutter_at_googlemail.com>
Date: Wed, 4 Jun 2008 00:22:57 -0700 (PDT)
Message-ID: <bc833cac-7143-4de0-843c-997a2f365c53@2g2000hsn.googlegroups.com>


On Jun 3, 4:38 pm, sati_82 <krys_..._at_interia.pl> wrote:
> On 3 Cze, 09:18, Robert Klemme <shortcut..._at_googlemail.com> wrote:
>
> > Hmmm, are user errors so frequent that you need a mechanism for this?
> > It seems, you might either need to better educate your users or
> > control access in a way that no damage can be done. Depending on the
> > application at hand and the nature of errors I can think of a number
> > of ways to do this, e.g. constraints, staging area where changes are
> > entered and are only propagated to the main data if a second user
> > confirms them.
>
> Thanks for your reply. Unfortunately neither user education nor DB
> constraints
> can be applied in such situation. The rollback mechanism is suppose to
> protect from a invalid data loading process. In the system users are
> able to load
> data from a file into DB. The most frequent error is that the file can
> include
> incorrect data (although the file structure and relations between data
> can be correct).

How do you determine that data is invalid? If there is an automatism, what stops you from loading data into a staging table and copy it only once it passes the check? You could also load the data into a staging table and write a procedure that copies it to the target table, does all checks and rolls back if errors are found. In every case, RMAN does not seem to be the appropriate tool for the job. My 0.02EUR.

Kind regards

robert Received on Wed Jun 04 2008 - 02:22:57 CDT

Original text of this message