Full weekly offline copy

fredric

ADSM.ORG Member
Joined
Aug 23, 2007
Messages
42
Reaction score
0
Points
0
PREDATAR Control23

Hi,

I'm trying to find out the best way to perform a full weekly offline (air gap) copy to tape.
To protect from Ransomware, we would like to have a full weekly offline copy to tape. That I'm looking for is simular to "
Copying hydrated data to tape storage for disaster recovery" Copying hydrated data to tape storage for disaster recovery - IBM Documentation . But it seems only do a full copy the first time, then incremental. I want to create a new DR-copy every week for at least 6 months. It's ok to only have a backup/week. If it is done by incr, the node data will be spread on many different tapes.
Or is it better to perform a full backup of all the "nodes" (VMware, Exchange, Oracle, Windows/Unix) to a separate mgmt-class/pool(tape)? I'm afraid that this way would generate high license fees, due to more active data (Back-end Usage License).

All ideas are welcome.

Thanks!

//Fredric
 
PREDATAR Control23

What would work the best is using "Retention sets to tape" introduced in 8.1.10: https://www.ibm.com/docs/en/spectru...etention-sets-tape-air-gapped-data-protection

This works for:
  • regular client backup
  • VMs
  • MS Exchange (starting at 8.1.12)
  • MS SQL (starting at 8.1.12)
How this works at a high level:
  • On the client, you continue to do your incremental backups forever, no need to do a full backup at all.
  • On the server, you have a retention set that's schedule to run weekly, what that will do is take all the Active versions at that time, which would be all the same files as if you had done a full, and the server will put that in a retention set, and you set that to copy it to tape.
For the clients you cannot use retention sets yet like DB2, Oracle, Domino, etc., the best thing is to do a weekly full to a dedicated storage pool just for the fulls. You would also need a dedicated copy tape storage pools. After your fulls are completed, you would do a BACKUP STGPOOL, from the primary pool to the copy pool, and then use DRM to send those off site.

Here's a few useful videos with demo:
 
PREDATAR Control23

Many thanks marclant for your respond, I will look into it in more detaile.

Regarding the setup of a separate storage pool for just the fulls backup (Oracle), I supose the storage pool can't be a container pool, since I must use backup stgpool later on.

//Fredric
 
PREDATAR Control23

Many thanks marclant for your respond, I will look into it in more detaile.

Regarding the setup of a separate storage pool for just the fulls backup (Oracle), I supose the storage pool can't be a container pool, since I must use backup stgpool later on.

//Fredric
It could be, it depends on your needs.

If you create a new container pool for these backups, then each wheel after your fulls, when you do the hydrated copy to tape, it would only copy these new fulls, since that would be the only thing new in that pool.

If you use an existing container pool, do your weekly fulls there. You would create your rule for the hydrated copy, along with a subrule to only specify the Oracle node(s). However, when you do your hydrated copy to tape, it would take all the backups for the Oracle node(s) that have occurred since the last hydrated copy to tape.

If you create a new filepool (not diskpool) to hold the fulls, it's like with a new container pool, the backup stgpool will only copy your new fulls, since it will be the only new data in the pool since the last backup stgpool.
 
PREDATAR Control23

Thanks for your respond.

I noticed that I can't add the nodes with Oracle data backed up with rman, only the Oracle server-node. Is there another way to perform the full Oracle backup that can be picked up by the retention rule? Backup rman files to a nfs-share or dump/export the databases.
 
PREDATAR Control23

Thanks for your respond.

I noticed that I can't add the nodes with Oracle data backed up with rman, only the Oracle server-node. Is there another way to perform the full Oracle backup that can be picked up by the retention rule? Backup rman files to a nfs-share or dump/export the databases.
It's not supported at this time, as I mentioned earlier, retention sets are only supported with:
  • regular client backup
  • VMs
  • MS Exchange (starting at 8.1.12)
  • MS SQL (starting at 8.1.12)
So, you will have to use a method other than retention sets for nodes that are not of these types.
 
PREDATAR Control23

Ok, you are right, I forgot that you mentioned it earlier.
Will the data be treated as active data? If I store the backup data on tape for long time, will it affect the license (TB back-end)?

Thanks!
 
PREDATAR Control23

Ok, you are right, I forgot that you mentioned it earlier.
Will the data be treated as active data? If I store the backup data on tape for long time, will it affect the license (TB back-end)?

Thanks!
The data will be a mix of active and inactive. If you create a retention today and copy it to tape, it would be active. But when you repeat this process next month, the new retention set will be all active, the previous one will be a mix of active and inactive because some files will have changed, some will still be the same.

As far as how that's counted towards licensing, I have no idea, probably best to ask your IBM rep for that.
 
PREDATAR Control23

Ok, it makes sense.
It will probably increase the need for more licenses over time, I will keep an eye on the license number to see how it change.
 
PREDATAR Control23

Should I perform a weekly full VM backups (IFFull) to speed up restores from tape or doesn't it matter? Will the retention set create a full synthetic backup or will it backup the full + last incr (ifi) to tape? The tests I have performed so far indicate that it takes longer time to restore if I don't perform a full backup before the retention set starts.
 
PREDATAR Control23

Update.
Due to high SUR archive licensing I don't use retention sets anymore. When you take VM image backups, they become inactive at the second backup, which increase the need of licenses. Instead I backup the deduppool to a copypool of tapes, which I bring offsite. The data is hydrated on the tapes, so I can restore directly from them.
Keep in mind to store the VM CTL files in a seperate diskpool. If needed, begin to restore it back from the copypool (restore stg) to the diskpool before restoring from tape.
 
Top