Hi all,
I am trying to replicate my production VMs to my offsite Datacenter using vSphere Replication. Currently I'm on vSphere 5.1 (vCenter, VR, ESXi) and the VMs that I'm replicating are running Windows 2008 R2 and MSSQL 2005 SP4.
My issue is that whenever the replication starts I experience a significant degradation of performance on the replicated VMs. The error log shows
I/O is frozen on database model. No user action is required. However, if I/O is not resumed promptly, you could cancel the backup.
followed by
Sqllib error: OLEDB Error encountered calling ICommandText::Execute. hr = 0x80040e14. SQLSTATE: 42000, Native Error: 3013
Error state: 1, Severity: 16
Source: Microsoft SQL Native Client
Error message: BACKUP DATABASE is terminating abnormally.
SQLSTATE: 42000, Native Error: 3271
Error state: 1, Severity: 16
Source: Microsoft SQL Native Client
Error message: A nonrecoverable I/O error occurred on file "{3E366AB8-8407-4E31-8A30-2AEE44C49E22}4:" 995(error not found).
and then followed by
I/O was resumed on database model. No user action is required.
This occurs for each and ever database that is hosted on the server which poses a significant problem for me and makes it hard to justify an upgrade to an SRM solution.
I read up on many probable solutions which blame the MSSQL Snapsot Writer and generally tell me to point the finger at Microsoft and not at VMware. None of the solutions seem to work for me.
So I'm turning to you good people as I cannot comprehend why such an I/O lock up would happen with vSphere Replication snapshots, but never occur if the VM is snapshot using the standard method.
I do understand that VR snaps and ordinary snaps are alike, however there is a significant difference in what's going on in the background.
Any suggestions are appreciated and I would be very grateful if a solution is at hand. I do believe that I'm not alone with such a problem.
Best regards.
