MinCpu

MinCpu is a term used in some resource-management contexts to denote a configurable minimum amount of central processing unit (CPU) resources allocated to a workload. It is not a universally standardized field, but appears in various manifests, scripts, or controllers to guarantee a baseline level of CPU performance for containers, virtual machines, or functions. The purpose is to prevent CPU starvation during contention and to improve predictability of latency-critical tasks.

In practice, CPU resources are typically expressed in cores or millicores. A minCpu value would represent a lower bound on the CPU resources that the scheduler should allocate, potentially alongside a maximum or limit. It is commonly used together with reservations or guaranteed-quality-of-service concepts to ensure a stable baseline, while still allowing for scaling under higher demand.

Implementation can vary by platform. In Kubernetes, the canonical mechanism uses CPU requests and limits; there is no built-in minCpu field, but some custom controllers or manifests may expose a minCpu to approximate a guaranteed baseline, for example by setting requests equal to a specified floor or by using a specialized scheduler feature. In virtualization and cloud contexts, reserving a number of vCPUs for a virtual machine or a function can serve as a practical minCpu, ensuring that those resources remain available even when the host is busy.

Considerations include the potential for resource underutilization if the minimum is set too high, and scheduling complexity or over-provisioning if set too low. See also CPU allocation, resource quotas, cgroups, and QoS in scheduling.