You can actually turn off the VM quiescing which is the root problem here. Snapshot backups on a multi-tiered application like JDE are almost guaranteed to cause problems since the DB server will often freeze for a moment during the snapshot process.
Some backup software can be set to not quiesce during snapshot. If that is not an option then you can refer to this VMWare KB document:
http://kb.vmware.com/selfservice/mi...nguage=en_US&cmd=displayKC&externalId=2044169
Note that I only do this step below mentioned in the document:
Disable VSS application quiescing using VMware Tools
========================================
Open the C:\ProgramData\VMware\VMware Tools\Tools.conf file in a text editor, such as Notepad. If the file does not exist, create it.
Add these lines to the file:
[vmbackup]
vss.disableAppQuiescing = true
Save and close the file.
Restart the VMware Tools Service for the changes to take effect.
Click Start > Run, type services.msc, and click OK.
Right-click the VMware Tools Service and click Restart.
The KB article states that you just need to restart the VM Tools service after making the change. I have found that sometimes a reboot is needed.
The document also describes creating custom quiescing scripts which under Windows use the VSS service to backup open files. I have not used this feature since there are known issues with VSS and high I/O applications like SQL server which means that even with the custom quiesce script you still get the DB freeze/disconnect.
The key point to keep in mind is that just using the vss.disableAppQuiescing by itself will make the snapshot backups only "crash consistent". This means the snapshot will pick up the state of the file system while it is still active which is similar to the state of the system if the power is cut off. Since I rely on database agent-based backups and DB log shipping technology for the database my primary data corruption concern is dealt with. With a crash consistent backup it is possible that you could have some corrupt files on a restore. To mitigate a bit against this I aim for a cold backup of VM hosts (where I have disabled quiescing) at least once a month.
For a discussion on why crash-consistent is probably good enough for my scenario see:
http://www.altaro.com/hyper-v/vss-c...plication-consistent-vss-backups-post-1-of-2/
