cgroups v2
Control Groups version 2 — the Linux kernel's mechanism for limiting and measuring how much CPU, memory, and I/O a group of processes may use. One of the two technologies (with namespaces) that containers are built on: cgroups govern how much, namespaces govern what you see.
Mechanic: processes are placed in a hierarchy of cgroups, each with limits and weights for resources. When you set a CPU/memory limit on a container it's translated into cgroup settings; the kernel enforces them (throttles CPU, OOM-kills on a memory overrun). v2 (which replaced the messy v1) has a unified hierarchy (all resource types in the same tree, not separate ones), and better features — especially PSI (Pressure Stall Information) which measures how much processes stall due to resource shortage (powerful for detecting overload). Kubernetes and modern container runtimes now require cgroups v2. It's the "muscle" behind resource requests/limits. Distinct from namespaces (isolation/visibility). One of the two foundations of the Linux container. Related to namespaces (Linux) and resource requests/limits.