VIRTUAL CPU (vCPU)
Virtual machines are allocated one vCPU each by default. If the physical host has multiple CPU cores at its disposal, however, then a CPU scheduler assigns execution contexts and the vCPU essentially becomes a series of time slots on logical processors. In this regard some experts treat vCPU not as a separate CPU but a share of the time the core of the processer is used.
Because processing time is billable, it is important for an administrator to understand how his cloud provider documents vCPU usage in an invoice. It is also important for the administrator to realize that adding more vCPUs will not automatically improve performance. This is because as the number of vCPUs goes up, it becomes more difficult for the scheduler to coordinate time slots on the physical CPUs, and the wait time can degrade performance.
In VMware, vCPUs are part of the symmetric multi-processing (SMP) multi-threaded compute model. SMP also allows threads to be split across multiple physical or logical cores to improve performance of more parallel virtualized tasks. vCPUs permit multitasking to be performed sequentially in a multi-core environment.
