Cluster SVM scoped and data SVM scoped NDMP backups for Commvault backups

[sanmonkey]

We are currently using data SVM scoped NDMP backups using Commvault.  We have about 6 data SVMs in an 8 node cluster.  For reasons I don't recall, years ago we chose to configure NDMP at the data SVM level.  Each data SVM is its own client computer in CV with appropriate agents (NDMP, Linux, or Windows) and subclients configured.  We now are testing SVM-DR (which seems pretty cool) to a peered cluster in another datacenter.  It has become apparent that CV needs to be configured at the cluster admin SVM level in order to do Intellisnap and backup the data when we fail over to the DR cluster.   My question is, can NDMP be configured at both the cluster admin SVM level and at the data SVM levels and backup just the SVM-DR volumes via the cluster or is it an all or nothing sort of thing where I need to re-configure everything to backup that way?   I'd like to backup just the SVM-DR via the cluster (for now) and leave the others where they.  

[Sergey_Osipov]

Reconfiguring backups to admin SVM -> means it is a brand new ndmp server for a backup application

Meaning it won't know the history of backups made for ndmp within Data SVM

and the same opposite

[sanmonkey]

The data SVM configured for SVM-DR is still in testing and hasn't gone into production yet, so in this case there's nothing to lose.  Once we're done hashing out the details, we'll be re-hosting some some volumes from another data SVM that we want to protect with SVM-DR and we know those initial backups will be fulls.  We would disable current subclient and "should" be able to access original backups prior to re-hosting since that will be a different subclient under a different SVM.  We're trying to get NetApp and Commvault together in a call to hash this out.  I wish we'd have done cluster SVM scoped NDMP from the beginning, but hindsight is 20/20.