Re: DataGuard Standby Database usage

From: Dan Norris <>
Date: Sun, 01 Jun 2008 20:36:42 -0500
Message-ID: <>


Thanks, Asif. I saw Daniel Koehne's email just after I sent my original
note and realized that I'd forgotten about that "workaround". I haven't
looked closely at the 11g snapshot standby feature, but I'd have to
guess it's built on top of the 10g flashback database features as
employed in the procedure you outlined. 

At any rate, you're right--I've just had 11g new features on the brain a lot lately, so it was the first to pop in to my head.


Asif Momen wrote:
Hi Dan,

>> It won't help you on 10g, but this is exactly what the new snapshot standby features in
>>11g are for.

Even in Oracle 10g, you can simulate 11g snapshot standby feature on a physical standby database. This is achieved by:

1) Removing Physical Standby database from recovery mode
2) mount data
3) create a restore point
4) open physical standby database in READ-WRITE mode
5) once testing is complete, close database
6) flashback database to restore point
7) open your standby database and place your physical standby database back in recovery mode

I did this couple of times on our 10g databases, if you need step-by-step instructions  then just drop a mail.


Asif Momen

--- On Fri, 5/30/08, Dan Norris <> wrote:
From: Dan Norris <>
Subject: Re: DataGuard Standby Database usage
Cc:,, "ORACLE-L Freelists" <>
Date: Friday, May 30, 2008, 9:35 AM

It won't help you on 10g, but this is exactly what the new snapshot standby features in 11g are for. On 10g and before, the only effective (which is still pretty rare) approach I've seen is to use storage snapshots to take a (cold) snapshot of the standby database before activating it. When the testing is done, the activated standby is dumped and replaced with the snapshot copy made prior to activation.

YMMV, not all environments have storage snapshot capabilites. You can obviously make a backup of the standby prior to activation, but that requires disk space and time as well.


Howard Latham wrote:
We have the same problem our standby takes abour 12 hours to build and when opened for a test thats it - its not a standby any more
so sucessful test then destroy it to build another untested one. Flashback standby anyone?

2008/5/30 Joerg Jost <>:
Am Freitag, den 30.05.2008, 06:51 -0400 schrieb Terrian, Thomas J Mr CTR
> Oracle 10g, HPUX, Physical DataGuard, RMAN backups of Primary and
> Standby.
> What would be the best way to test our DataGuarded copy if we want to
> start it up and applying changes to it?  We don't want those changes to
> come back to the primary.
> Can we start it up as a regular database but then when we are done use
> our RMAN backup of it to restore it to the point were it was still a
> standby database?  Will the primary database have any problems with
> this?
> Anyone done anything like this before?

Hi Tom,

if we open a standby database for writing, we build afterwards a
complete new standby database. Using rman we can do it anytime, even
during working hours, so we don't try anything spezial.



Howard A. Latham

-- Received on Sun Jun 01 2008 - 20:36:42 CDT

Original text of this message