|
Virtual hard disks are typically files created within the host file system. They can either have a fixed size (that is, disk space is completely allocated when they're created), or expand based on actual storage requirements (thereby saving disk space). This system provides a lot of flexibility in allowing you to move virtual hard disks between host computers. The potential downside is that it takes system resources to map between virtual read/write operations and their physical file I/O (input/output) counterparts. On the upside, virtual hard disks can be assembled into hierarchies and can be rolled back to specific points in time.
Mapping directly to physical disks is often done for the purpose of performing a physical-to-virtual (P2V) conversion, or where overall performance is most important. The drawbacks include more difficult management (since you'll have to plan out your partitions), and less efficient use of disk space.
Overall, I recommend you stick with virtual hard disks for most applications and services. Direct physical mapping might be appropriate for specific workloads in which disk performance is important, or you have requirements such as direct access to network-based storage.
|
