257 Posts

July 16th, 2009 07:00

Yes indeed.

If the same (target) clone is used, it will only be an incremental update (track copy), which will of course be much faster.

James.

22 Posts

July 16th, 2009 07:00

James,

That's great news, it took 4 hours to create the clone yesterday.

Also, is there a document anywhere to I can view to give me a good general idea of how other customers use RM with Oracle? The purpose of this clone is going to be to offload a hotbackup that we currently run every night on our production database.

We're also wanting to create a snap shot every 4 hours that we could mount to a different host to combat user error. It would be nice to know what others are doing to take full advantage of RM with Oracle.

Thanks,
Ben

257 Posts

July 16th, 2009 07:00

Hello Ben

Yes indeed, so the terminology for sync'ing / re-syncing is called a job.

Appset = what you want to replicate/clone (say oracle)
Job = how you want to replicate/clone (eg snaps, clones, SRDF, etc)

So, when you re-run the same job you previously created, the existing clone will be unmounted from the mount (target) host and the same clone device (unless you have added additional clone devices) will be used to re-sync.


Hope this helps

James.

22 Posts

July 16th, 2009 07:00

Hi James,

Thanks for the reply. So when I execute this same job again, it's going to automatically know what blocks have changed and only sync those blocks?

This particular job is cloning an Oracle database with consistency method of online using hot backup mode.

Thanks,
Ben

22 Posts

July 16th, 2009 08:00

One more question.

I just simulated the same clone job again and when it got to the part where it mounts the vg/luns back to the target host, it failed on the mount due to those mount points already being mounted. Do I need to manually ( or within a pre-script ) unmount that volume group. I would have thought the sync would fail because it couldn't modify luns that were mounted to a host.

2.2K Posts

July 16th, 2009 09:00

If you run a simultation of an existing job (clone in your case) that is already mounted to the host, then I think that is the expected behaviour. The job has already been run once and the devices are currently mounted to your mount host, so that part of the simulation would fail. I have had this same thing happen to me when troubleshooting an existing job that was mounted to a host already.

There are some good white papers on the use of RM, one at least concerning Oracle. Go to:

Support > Technical Documentation and Advisories > Software ~ P-R ~ Documentation > Replication Manager > White Papers

You should see the Oracle and RM white paper there.
No Events found!

Top