Snapmirror Certain VMDK's

Hi All,

We have some replication issues due to the amount of data we need to replicate.  The question i have is can put all DATA drives from servers in one DATASTORE and snapmirror that datastore?  I would like to keep the 'C" drives of all the servers seperate from the data drives and NOT replicate them.  What i'm not sure of is how VSC takes the snapshot.  I'm just trying to limit the amount of data that we replicate.  Any thoughts or comments are welcome.  Thanks!


Re: Snapmirror Certain VMDK's

why not mirror the current servers to another volume using vsm after that quiesce and break the mirror and then remove the d and e drives etc from the original and then add back the data drive to the servers on the original volume and delete the old c drive vmdk and -flatvmdk, the vsm should still work but will takes snapshots on the new volume and old server volume too for the c drive as the vmx will be updated with the new paths, worth testing first though with one machine.

Re: Snapmirror Certain VMDK's

NetApp snapshot is volume level.

>>The question i have is can put all DATA drives from servers in one DATASTORE and snapmirror that datastore 

This is possible, effectively your second volume (datastore) will be backed up.

I am not sure whether VSC allows you to backup a datastore without any VM's (but only vmdks present in it... - you can try this out)

But as you are interested in backing up only data drives and not the VM's (assuming that VM consistent snapshot is not required) you can do this from the filer itself. snapmirror the

data drive datastore volume. (you will have vmdks saved on the secondary).


Re: Snapmirror Certain VMDK's

First of all guys I am a rookie when it comes to the Netapp.  I know enough to be dangerous.  We had a vendor setup our original Snapmirror and snap vault for us.  Here were my thoughts of how i think it should work out.

Move existing VM's to a different datastore placing C drives in datastore1 and the data drives in datastore2

Then setup a snapmirror relationship for datastore2 and just have that volume replicated.  Datastore1 would not get replicated. 

Is that doable?

Re: Snapmirror Certain VMDK's

This is verymuch doable.

But what do you have in data drives? Don't tell me you have databases (SQL, Oracle) .....

If they are just flat files, then its okay.... because during the snapshot / backup i/o wont be quiesced as in VMware consistant snapshot. so it will be like a flat file backup.


Re: Snapmirror Certain VMDK's

The data is mostly SQL data.  Not active databases but client data.  One server has client financial data that needs to be retained for a long period and the other server has development SQL projects for clients that needs to be retained for a longer period as well.  That is why i need to cut down on the amount of data that is replicated. 

Re: Snapmirror Certain VMDK's

what i am saying is create another blank volume and mirror the whole of the existing server volume and add the d drives back to the servers from the new mirror and then you can resynch the old server destination with the one you have mirrored after you have removed the c drive vmdk and break the old snapmirror relationships, if you are using vsm this is block level replication if you need to backup databases etc you should have snapmanager or some other backup product.

But what you are trying to do is totally normally some people have the swap files files on different volumes.