We are in the beginning phases of a Recover Point Installation. Our Windows people have a few questions regarding Hyper-V VMs running on
Windows 2012, and Windows 2012 R2.
When a lun is being replicated and replication is stopped for some reason, planned or unplanned, what happens if there are open files on the VM or database(SQL)?
When snaps are being taken, what is best to use for VMs, crash consistency, or application consistency option?
Thanks in advance!
If replication is stopped the VM and associated database contines to function normally. One of the underlying benefits of RP is that is '"out of band", meaning that the interception of the write I/Os and any impact to this process does not interfere with the application/database.
Depending on the nature of the replication stoppage (pasue vs. error), RecoverPoint may be able to mark the mark the writes as metadata in the CG source copy journal and then when the issue is resolved read this the actual data from the array and replicate it.
Regarding your second question, it depends. Crash consistency vs. application consistency simply comes down to a discussion relating to RTO (Recovery Time Objective). A crash consistency snapshot will take longer to recover than an application consistent snapshot because the database does not have to roll it logs forward and reach a consistent PiT. This takes times compared to the application consistent snapshot where the database is clean and no roll is required.
I hope this helps.