With metrocluster, the remote site (Site B) contains disks that are written to simultaneously with the local drives (Site A), but the remote cluster (Site B) does not enable access to them unless switchover occurs. There is an option to enable read from both local (Site A) and remote (Site B) drives, but it is from the Site A controller cluster that the reads are enabled.
If I'm guessing your motivation for this question - it is either to verify the remote copy, or to reduce intersite bandwidth/latency. There is no need to verify the remote copy, it is taken care of by the RAID subsystem - this is not a "backup", it is active-active, with all of our raid scrubs and on-disk verifications that occur. The monitoring of metrocluster only requires monitoring of the Cluster Replication Service status inside of ONTAP, and if implemented, the MetroCluster Tie Breaker in a third site (Site C).
If latency or lack of bandwidth between sites is an issue, ensure this is dealt with before Metrocluster is implemented, as there are very low tolerances for latency or bandwidth limitations.
Hope this helps - and if I've misguessed why you are interested in this functionality, please feel free to ask any follow up questions or provide details to help me understand your use-case.
thanks for your reply. Yes my motivation is to see data inside remote Site B, for example, we are an email hosting provider, we provide with snapshot and snapmirror ability to view your mailbox in the "past" (like a time machine) and read this data only from replica FAS.
I have another question, is there any configuration with NetApp where, with 2 FAS, we can have an NFS volume replicate and writable from both?
Is the SiteB in a different location and this is provided for geographic redundancy, or do you do this to reduce read workload on the primary system, or is just because this is done currently?
Multi-site R/W is a very difficult problem to solve computationally. I think the "storage raid-options modify -raid.mirror_read_plex_pref alternate" option to spread read workload between both sites is the best choice.