Changing Storage Unit within an NBU Accelerated Policy

[Fred2010]

Hi,

We have several Netbackup Accelerated Policies that backup to a Netbackup 5230 MSDP pool

Unfortunately, the backup unit on the NBU appliance is too full, so we have to change the policy to use another Storage unit (HP StoreOnce Catalyst) capable of Accelerated backups...

I understand that on VMware the Change Block Tracking map that is in use needs to be regenerated (As it was build up some time ago)

Is forcing an Accelerator Forced Rescan sufficient to get the CBT log regenerated, so that the first backup to the HP Storage gets ALL the blocks instead of just the changed blocks since last backup on the MSDP Storage?

If not, how can this be done best?

Thanks for your help!

[RiaanBadenhorst]

There is no dedupe between these devices so it would send all the data if you backup to a new destination.

[Fred2010]

Hi Riaan,

I understand that there is no exchange of data between the two storage units, but my question was regarding the CBT log that is present on VMware:

The CBT map was created and populated when the MSDP was in use, and was only sending out changed blocks when it was backing up to the MSDP unit.

I then changed over the Storage unit to HP StoreOnce and I assume the CBT map on VMware doesn't get rebuild automatically and still only sends over the changed blocks I assume???

Am I correct if I say that the backup on the HP Storage is incomplete (As it only received the changed blocks) or is the CBT reset, so that the first backup to the HP Storage unit is actually a full backup based on a fresh and empty CBT map?

Or do I need to do an Accelerated Forced Rescan to rebuild the CBT map on VMware... If not, what is the best way to get the CBT map reset again?

Thanks!

[Nicolai]

I think Netbackup will do a automatic full when Netbackup detect you have changed the storage unit, it might call it "incremental" but the detailed text will say its doing a full backup. 

Netbackup does a full upon storage unit change in order to secure all data needed to do a full backup is residing on the same storage unit.

So it should be impossible to have split one full backup across storage units (e.g uniq block for full backup across two storage units).

By the way, CBT is a VMware part not related to Netbackup as such. NBU uses the CBT but so does other applications also.

[RiaanBadenhorst]

Hi

I don't have any document to support this but I'm sure that it will understand that a different storage server is being used. The storage server (deduplication device) is responsible for the creation of the synthetic from the bits that are sent to it. As you do not have an image on the new storage that storage server would not have any reference of the client and that should ensure a full is taken with out the need to run forced rescan.

Maybe we can get a confirmation from a engineer/backline.

[Fred2010]

Hi Nicolai,

Thanks for your answer. Indeed, the CBT is part of VMware and Netbackup uses it to 'see' which blocks have changed to be backed up...

My doubt however is with the CBT: Is it reset somehow by the changed Storage Unit?

Or is Netbackup smart enough (As Riaan suggests below) and sees it has not all the bits on the new storage unit and makes a full backup of all blocks anyway.

My problem lies with the CBT: What mechanism resets it, or doesn't it need any resetting?

Thanks!

[Nicolai · Accepted solution]

I found thease links and think Fred will find them instresting:

How to reset Change Block Tracking (CBT) for VMware Backups

http://www.symantec.com/docs/TECH197311

NBU and the CBT issue for vmware

https://www-secure.symantec.com/connect/forums/nbu-and-cbt-issue-vmware