VCB recursion

[H_a_r_m_o_n_t]

I have VCB which is installed inside of VM. I want to perform image-level backups of VMs. When one perform image-level backup entire VMs one by one are copied into the folder inside of VCB VM. How can I copy VCB VM into VCB VM? It obviously will not fit.

[depping]

You will need to attach an additional disk to the VM which can contain the vm images. This could be an RDM or a VMDK, that's up to you. Keep in mind that when VCB does a full image dump it normally doesn't copy each block, it only copies the blocks that contain data. So in that case it might be useful to zero-out the VM before you start VCB'ing. I wrote an article on this: http://www.yellow-bricks.com/2008/01/04/vmware-consolidated-backup-and-deleted-files/

Duncan

Blogging: http://www.yellow-bricks.com

If you find this information useful, please award points for "correct" or "helpful".

[H_a_r_m_o_n_t]

You will need to attach an additional disk to the VM which can contain the vm images.

You mean that if I add separate .vmdk specifically for storing VM images I can later exclude it from backup? But how? As far as I know it is possible only while performing file-level backups of VMs not image-level.

[H_a_r_m_o_n_t]

Could somebody further clarify this?

[Craig_Baltzer · Accepted solution]

The only place this is an issue is if you're trying to backup a VM running VCB to itself (i.e. you're backing up the backup server onto itself). So...

1. When a VCB backup starts a snapshot is opened. The backup will now run against data that is "consistent" as of the start of the backup, and any changes that occur as the backup is running are ignored. So the fact that VCB is writing to disk in the VM won't get into a "recursive loop" and run itself out of disk space by re-backing up data over and over again.

2. As Duncan mentioned VCB only backs up disk blocks that contain data, so if you started the backup with an "empty" VMDK file that you were going to to use to hold the VCB mount point and backup files, then there would be very to back up (i.e. if it was a 60GB VMDK file then the backup would only contain the blocks from the VMDK file that actually had data, not the entire 60GB). So if the target disk is empty when the backup starts it won't be "filled up" by trying to make a full copy of itself

So just as a test I had VCB installed on a VM with a 12GB C drive (OS) and a 65GB F drive (mount point for VCB). The C drive had a bit over 3GB of space in use, and drive F was bascially empty. Using vcbmounter I backed up the VCB VM, and when it completed I had 3.23GB stored on the F: drive.