Overcommit is a feature of the hypervisor that allows virtual machines to be allocated more resources than they have on the physical server. The concept is closely related to the term overprovisioning. Initially there is overprovisioning (allocating more resources than available) and only then overcommit (using more resources than available).
CPU overcommit is using more CPU resources than are available.
Correct operation at high CPU overcommit rates is the basis for cost-effective virtualization solutions in the software-defined computing layer. The technology helps reduce costs per virtual machine by dynamically reallocating power. Reallocating free CPU cycles helps increase resources and allows more virtual machines to be scheduled than would be available without correct CPU overcommit conditions. CPU overcommit is especially valuable for workloads where peak bursts alternate with periods of slack.
High CPU overcommit is the reason for the high cost efficiency in the SDC layer.
MEMORY overcommit is an overcommit memory allocation.
It allows virtual machines to be allocated more memory than they have on the physical server. This is achieved by the fact that virtual machines do not always use all of the memory assigned to them. The disadvantage of MEMORY overcommit is in the form of a swap file, the latency of access to which is orders of magnitude higher than to memory, which can cause a drastic performance degradation with an extremely negative effect on the guest operating system.