We are currently running about 70 VM's. And we have a 4 node 2012 Hyper-V cluster. Right now the CSV that they are on is about 10 TB. However we are not currently looking into splitting the guest VM's up by OS into separate volumes. Also creating Separate volumes for each other role the. Putting all the SQL Drives into one volume. All the SQL Tlogs into one volume and so on. Heard that is how to get the best de-dupe out of the systems.