Performance and capacity management in the physical server environment is a relatively straightforward process, in contrast to the world of server virtualization. Key performance metrics are collected to understand the peak resource requirements of a workload, and enough capacity is provisioned in the physical server in order to support this, along with some additional contingency. This approach has lead to lots of wasted capacity: on average, across typical physical server farms, utilization levels range from 5-15%. The cost implications of this are significant.
Virtualization has proved to be a great solution to improve these economics because it allows multiple workloads to reside on the same physical hardware, enabling significant physical server consolidation and resulting in significant ongoing capital and operational cost savings.