In a disaster recovery with stretched storage, the failover command must isolate devices at the recovery site.
If some hosts at the protected site are still operational and continue running virtual machines when you initiate a disaster recovery, Site Recovery Manager cannot power on the corresponding virtual machines at the recovery site due to file locks. If the storage array isolates the devices at the recovery site, the ESX hosts at the recovery site can break the necessary locks and power on the virtual machines.
Site Recovery Manager must use
isolation="true" in the failover SRA command for the stretched devices that were not deactivated at the protected site.
If there are VMs running at the recovery site from the same device, and the recovery site ESXi is mounting the storage from the protected site, during isolation there is a risk of failing write operations. It is recommended that all VMs on stretched storage are running on the protected site.
Implementation details of isolation for stretched storage are specific to array vendors. Some array vendors might make the devices inaccessible at the protected site after running the failover SRA command with isolation. Some array vendors might break the communication between source and target site for that particular device.