2016-03-15 09:26 AM
We have just begun a 7Mode migration from a FAS3140 that runs both cifs and NFS to a new FAS2520.
The new primary Data ONTAP 8.3.1 Cluster is hosting a CIFS SVM on our LAN and an NFS SVM on our isolated storage network.
We also have a new dedicated DR Cluster for replication that will be moving to our CoLo once all the data has been replicated to the new cluster.
We have created a Peer Relationship over the Default IPSpace / LAN Subnet with the DR 8.3.1 Cluster and have extablished Data Protection SnapMirrors for several volumes located on the cifs SVM.
The NFS SVM is in a different NFS_IPSpace we created and along with a new NFS_Subnet since it is on an isolated storage network not accessable over the LAN.
We are currently running NetApp Virtual Storage Console 4.2.2 since our ESXi hosts are still running vSphere 5.1 with a vCenter 5.1 server.
We have added the a new Cluster DataStore to the vCenter 5.1 server and ESXi 5.1 hosts and have vMotioned the virtual machines to the new 8.3.1 Cluster DataStore.
We will be replacing the vCenter 5.1 server with a vCenter 6.0 server running VSC 6.2 and migrating the ESXi 5.1 hosts to the new vCenter 6 server in the near future.
Now we would like to create a SnapMirror Relationship for the NFS SVM and/or the datastore volume hosting the VMware vSphere virtual machines.
The problem is since this SVM is in a differnent IPspace/Subnet, we are not sure how to create the SnapMirror relationship.
We do not currently use VMware vCenter Site Recovery Manager but would like to make sure we allow for it in the future if desired.
Should we create an SVM SnapMirror Relationship for the NFS SVM?
If so, since it is in a diffent IPSpace/Subnet will we be able to create an SVM Peer Relationship even though we currently have a Cluster Peer Relation skip over the LAN?
Or should we creat a volume SnapMirror Relationship and craete a new NFS SVM Destination in the NFS_IPspace?
2016-03-21 05:45 PM
Peer relationships are created using the intercluster LIFs, which exist at the cluster level, not at the SVM level. It should be fine to create SnapMirror relationships from your primary cluster's NFS SVM to your secondary cluster even with the differing IP spaces and subnets. Just make sure that the destination SVM has access to the NFS network so that way if you need to failover the storage will still be accessible to the vSphere hosts.
Also, assuming you have two clusters which are sharing the same network connectivity, you may want to look into SVM DR (page 113).