Understand the impact of your virtualization vendor feature set! : Virtualization technologies provide a ton of useful features, such as dynamically moving virtual machines of your SharePoint farm to another physical host. A common problem during such a move, I have experienced, in virtualized environments is the corruption of index propagation to the query servers when the index or query servers have been moved. Therefore be careful when pushing virtual servers to different host machines. Test these features before using them in your production environment.
Don’t let governance slip in your virtualized SharePoint environment: Apply the same rigour and change control processes to your virtual farm as you would a physical farm. Review and ensure change control processes are in place and followed to protect your virtual SharePoint farm.
Restoring snapshots: Microsoft does not support snapshotting of individual SharePoint servers as state stored in a snapshot may be incompatible with the rest of the farm. (Think of effect of snapshotted timer jobs running on each SharePoint server.)
Beware of over subscribing host servers: It is easy to underestimate how load will increase over time on each of your guest machines and cumulatively on each of your host servers. This problem is often caused by squeezing virtual servers in temporarily to meet short term business objectives when no budget is available to purchase hardware. This is dangerous, not only for SharePoint, but for all the virtual servers running in your farm.
Host is a single point of failure: A host is a single point of failure. If your host server dies, all virtual machines will fail as well. To overcome this, first load balance your vm’s across hosts, (Place virtual web query servers on each host.) Additionally use built in Windows virtual machine clustering features to ensure failover of virtual machines to the passive host server.