Consolidated backup

[mrodrigu]

I am trying to install Vmware Consolidated backup. I've read the documentation but still is not so clear to me.

As I understand, I need to install the Vmware consolidated backup framework apart from the ESX but connected to same SAN.

Then, I need to give the same storage disk in which VM's are, to the vcb server.

After that, I can use mountvm command to mount the vm disk on the vcb server to backup the information to a tape.

Or I can vcbmounter to mount on the vcbserver all the files (vmdk,logs,vmx) from a specific VM.

So far is ok??

I dont understand why do I have to install backup clients on each VM and on the other hand, if I need to make the disks with vmkfstools -d thick option. If this is true and I have disks without this option, do I have to delete them?

The other question is if it is true that I can do the snapshot without interrupt any transaction to the SAN. Because on other systems it is recommended to stop the services.

Thanks very much for your help. I need to understand the architecture to do some tests here.

[kjb007]

Without VCB, if you want to backup your virtual machines, then you have to have clients on the service console to backup the disk files, or use the backup client in the virtual machines to perform the regular backups.

I believe the default disk type when you create a disk from vc, is zeroedthick, so you should be ok there.

For the snapshot question, vcb creates a snapshot before it creates a mount. This is usually ok, unless you have a high I/O application running on that, like SQL Server or Exchange. For those type of applications, it is a general good practice to stop the application using pre-backup scripts to stop the application, then create the snapshot, and then perform the backup.

The rest of your comments are good. You'll have to have VCB proxy server, and give it access to the same LUNs that ESX is using for the virtual machines. Make sure the vcb proxy server is configured to see the same LUN id's. You can't use physical mode Raw Device Map (RDM) LUNs.

-KjB

[mrodrigu]

Thanks very much for your answer.

zeroedthick means that the virtual disk can be accessed by more than one node concurrently??

That is by default ??

[kjb007]

Actually, the filesystem is what allows the concurrent access. Storing your files in a VMFS datastore, is what allows multiple ESX nodes to be able to access it. ESX will use SCSI reservations to lock and release locks on the LUN in order to access the files stored in that space. zeroedthick simply means that the entire disk space you asked for is allocated at create time (as opposed to thin, which does not fully allocate the disk at create time), and will be zeroed out as the space is used. eagerzeroedthick is slightly different in that the disk is zeroed immediately, instead of waiting for a later time.

-KjB

[mrodrigu]

if I need to do a cluster with one virtual disk to be accessed by two nodes, I need to create it with the "vmkfstools -d thick" command.

You say that this is not necessary?? This is by default??

[kjb007 · Accepted solution]

Is your question about Microsoft clustering? Then, you should create the disk with eagerzeroedthick or thick, as the cluster-in-a-box configuration states. It is not required if you are setting up regular VM's and/or a VMware DRS/HA cluster.

-KjB

[mrodrigu]

Ok. I understood. Thanks very much

[kjb007]

Don't forget to leave points for helpful/correct posts.

-KjB