We offer DR for our customers to our second datacenter. All VMs are replicated by synchronous storage replication to the second DC. Previously we had one SCVMM that would see both DCs and all Hyper-V clusters in it, but when we would test a DR for a customer, SCVMM would get very confused.
Procedure with 2012R2 was to snapshot the storage, mount the snap CSV on the DR cluster, scan the CSV for VM files, register those VM, connect to DR network (isolated) and boot the VMs. However, because of the ID inside of the VM files, SCVMM would get confused and see double VMs. In the production cluster the VM would then disappear from SCVMM until we finished the DR test and remove the VM from the DR cluster.
Also register the VMs would sometimes be difficult because we would read the XML file, search for the disks and register the VM and re-attach the disks. Problem is that the CSV could be "disk15" in Prod Cluster and "disk10" in DR cluster. That is very difficult to script.
Now with 2016 and the encrypted XML we have to look for different ways to do a DR and most importantly also a Test DR without interfering with production.
For safety we're looking at at least implementing a second VMM for DR, but we can't seem to find a reliable method to register the VMs and their disks, because the CSV volume would still have different names. We could make it very complicated by writing a text file in each volume, the says what the original name is and do mappings based on that. But I think that is very risky.
Any tips on how to do this are very welcome.