cgroups (control groups)
termIT supportcloud
The Linux kernel facility that limits, accounts for and isolates the resource use of a group of processes - processor time, memory, block input and output, network - and one of the two features that make containers possible.
Originally built at Google to stop workloads on shared machines from starving each other, it answers the accounting question that virtual machines answered with hardware emulation, at a fraction of the cost. Every memory limit in a container platform, every noisy-neighbour mitigation and every out-of-memory kill you have ever debugged in Kubernetes is cgroups doing its job - which is why the platform's behaviour is only fully explicable at the kernel level.
Also known as: control groups, cgroup v2, resource limits