Re: Duplicating a database AND maintaining it

On Sep 26, 10:36 am, Donkey Hot <s..._at_plc.is-a-geek.com> wrote:
> Sounds like replicating, right?
>
> But the case is this.
>
> We now have a single production database machine having 4 different SID's
> (instances) on it. Let's say the machine is called DB.
>
> My task is to create a new setup, which has machines DB1 and DB2, where DB1
> is a clone of DB, and DB2 is a physical standby replica of DB1.
>
> Data Guard and making the replica between DB1 and DB2 is clear to me, I
> have built and documented the process.
>
> Now I would like to create a copy of DB data to DB1 and DB2, and maintain
> the data flow somehow from DB to DB1, which then sends it to DB2.
>
> I can create the initial copy from a cold backup. But how to maintain the
> replica of DB in DB1, WITHOUT ANY CHANGES to DB's setup? DB generates
> Archive log all right, I could use it. But how?
>
> All the systems are 10g (10.2.0.2) running on Windows 2003 Server with
> identical hardware and hard disk structure.
>
> All the databases have same db_name across all machines, although
> unique_dbname will differ.
>
> Can I read in Archive logs from DB to DB1, and if yes, how?
>
> This arrangement between DB and DB1 is of course meant to be temporary.
> When the customer accepts the system, the DB is dropped off and DB1 will be
> the production server.
>
> Any pointers?

You could implement Streams between DB and DB1; this would replicate the data from databases on DB to those on DB1 and not clutter the DB1 environment with additional archivelogs as they would be 'mined' on the source server. Presumng you have no 'offending' data types in the databases this could work nicely to 'pump' transactions from DB to DB1, while the physical Data Guard configuration writes the archivelogs to DB2 for application, keeping everything 'in synch'.

David Fitzjarrell Received on Wed Sep 26 2007 - 11:11:44 CDT