split-mirror offline backup??

[Former Member]

Hello all,

When looking at the documentation for the DB and Oracle under SAP, I found the following texts:

_______________________________________________________________________

Split Command Scenario

You start BRBACKUP on the backup host using the following command:

brbackup -t offline_split

BRBACKUP performs the following steps:

...

1. Shuts down the database on the productive host, in order to guarantee the consistency of the

database for an offline backup.

2. Splits the mirror disks using split_cmd.

3. Restarts the database on the production host, so that it is again available for the SAP System.4. Backs up the mirror disks on the backup host.

5. If you have set resync_cmd, synchronizes the mirror disks with the originals.

The resynchronization process does not normally use the CPU on the production server.

For more information, see the detailed plan below.

________________________________________________________________________

My question was that when #3 takes place, one would still have to recover some kind of offline redo log or something because when the DB is started again, it is changing

I suppose I am trying to see why anyone would use this type of backup scenario ... I thought the split mirror was to be specifically used for collecting online backups

TIA

[Former Member]

OK then, thanks for the input!

[Former Member]

2. Splits the mirror disks using split_cmd.

You split your mirror disks, and have the full copy of your DB on this mirror disk , you start DB on original disks, but you do copy from

splited mirror disks

. I so understand the spelling.

Regards.

[Former Member]

I saw that one copies the data files from the disks that are splitted(?) off but I was wondering of what use is it? To save time?

When you do an on-line backup, at least you know that what you have been doing will be a part of the backup when it is completed. So, for example, if I do a client copy while an online backup takes place (I know it is not a good idea - just using it as an example), then I know that the DB changes will be a part of the backup.

With this offline setup, the client copy would not be a part of it and you would probably have to gather all of the offline redo logs to be sure you could recover the DB to the point where you did the client copy (since the full offline backup would not have it).

Im just trying to see what this would be useful for.

Edited by: Casey Harris on Feb 10, 2009 1:08 PM

[Former Member]

A split mirror offline backup is technically the same as any offline backup. You have a consistent backup of your database, of course you still CAN do a recovery, but don't have to.

but I was wondering of what use is it? To save time?

- to save time, your downtime is only the time to shutdown, split, and restart the db

- to reduce load on the original disks, this depends on the layout of your mirror

We still have systems where we still to offline split backups, but others where we only do online splits.

Regards, Michael

[Former Member]

>

> A split mirror offline backup is technically the same as any offline backup. You have a consistent backup of your database, of course you still CAN do a recovery, but don't have to.
> 
> 

but I was wondering of what use is it? To save time?

> 
> - to save time, your downtime is only the time to shutdown, split, and restart the db
>
> Regards, Michael

If you restart the DB, then (IMHO), the datafiles would change and there would be archived redo log generated. In the meantime, you are copying files with a lower SCN.

So, does this mean that you need to

-> do another backup?

-> or collect the archived redo logs?

What would you do?

[Former Member]

The files don't change on the mirror. Your backup has exactly the SCN of the time you closed your db. Of course all the changes after the restart will only be covered in the redo logs. But that is exactly the same with any offline backup.

I think the understanding problem you might have, is that two things happen in parallel. The offline backup is still running on the mirror, while the real system is already running again. The mirror will be resynced (online) at the end of the backup, while the system might already be running for hours again.

Now guess, what happens if your backup fails? If you for example want to import support packages, so you need that offline backup and already started applying packages?

We in our case always check the backup prior to that and the redologs needed, to have a fallback.

Regards, Michael