the SVM-DR is designed to continue serving data at the DR location should the source encounter a catastrophic failure sufficient enough to cause an outage.
If you require reverting back to a specific snapshot that exists on both SVMs, then simply restore from a local snapshot.
If you require to revert to a specific snapshot for restoration of the SVM from the destination, you can follow the below process to re-mirror the SVM to the origin (production) and restore from the snapshot locally.
Mirroring data volumes would be ideal, but that can depend on your need. It would be ideal to mirror data volumes in an SVM-DR configuration to enable the DR site to continue serving data in the above described catastrophic failure example.
if i understand right, the SVM will be only used for DR purpose. if we have to restore the data from snapshot of volumes from destination to the source, we have to snapmirror it back.
On source, we usually keep short term snapshots for instantly reocvery, and on destination we keep longer term snapshots. if SVM replications will be only used for DR purpose, should we also set up volume snapmirror to destination and keep snapvaults as longer term backup copies for in case any restores will be needed? Since the SVM replication already included the volume replications, will be volume replications duplicated?
I'm not entirely certain I understand the question.
SVM-DR includes the attached volumes to the destination site when created. no need to create the data volume relationships.
Ideally you would never encounter a DR scenario. But, in the event one is encountered, you would bring up the DR site SVM and it's volumes (assuming you have the preserve identity enabled, you would not need to change where clients are pointed).
You can reactivate the prod site through the following process. (reversing the mirror)