Glossary of Terms
A
Access control list (ACL)
A list attached to a file or object granting specific permissions to named users or groups, beyond the basic owner/group/others model.
API server
The Kubernetes control-plane component that exposes the cluster's REST API, validates requests and is the only component that talks to etcd.
ASLR
Address Space Layout Randomization — placing code, stack and heap at random addresses so memory-corruption exploits cannot predict where things are.
B
Bootloader
Small program started by firmware that loads the kernel (and initramfs) into memory and starts it, for example GRUB or systemd-boot.
C
cgroups
Control groups — a Linux kernel feature that limits and accounts for the CPU, memory, I/O and process count used by groups of processes.
CIS Benchmark
A free consensus configuration guide from the Center for Internet Security describing how to harden a specific OS, cloud or application.
cloud-init
The standard first-boot tool for cloud and VM images that applies user data: users, SSH keys, packages, files and commands.
Container
A set of processes isolated with namespaces, limited with cgroups and started from an image, sharing the host's kernel.
Container runtime
Software that creates and runs containers — high-level runtimes such as containerd and CRI-O manage images, low-level ones such as runc start processes.
Context switch
The kernel saving one thread's CPU state and restoring another's so they can share a CPU core.
Copy-on-write (COW)
Sharing memory pages or data blocks until someone writes, then copying only the changed part — used by fork(), snapshots and some file systems.
CRI / CNI / CSI
Kubernetes plug-in interfaces for container runtimes (CRI), pod networking (CNI) and storage drivers (CSI).
D
Deadlock
A state where two or more threads each wait for a resource held by another, so none can ever proceed.
Deployment
A Kubernetes object that keeps a set number of identical Pod replicas running and rolls out updates gradually.
Device driver
Kernel code that lets the operating system control a particular piece of hardware through a standard interface.
DMA
Direct Memory Access — hardware moving data between a device and memory without the CPU copying each byte.
E
eBPF
A Linux technology for running small verified programs inside the kernel, used for tracing, networking and security tools.
EEVDF
Earliest Eligible Virtual Deadline First — the Linux kernel's default CPU scheduler since version 6.6, replacing CFS.
Error budget
The amount of unreliability an SLO allows (100% minus the target), spent on releases and experiments.
etcd
The consistent, distributed key-value store that holds all Kubernetes cluster state.
F
File descriptor
A small integer a process uses to refer to an open file, pipe or socket, such as 0, 1 and 2 for standard input, output and error.
H
Hypervisor
Software that creates and runs virtual machines; Type 1 runs directly on hardware, Type 2 runs on top of a host OS.
I
Idempotency
The property that running an operation repeatedly gives the same result, essential for safe automation.
Infrastructure as code (IaC)
Defining servers, networks and settings in version-controlled files that tools apply automatically, such as Terraform, OpenTofu or Ansible.
Init system
The first user-space process (PID 1) that starts and supervises services; systemd on most Linux distributions, launchd on macOS.
Inode
A file-system record holding a file's metadata and data locations — everything except its name.
Interrupt
A signal from hardware (or software) that makes the CPU pause its current work and run a handler in the kernel.
IOMMU
A memory-management unit for devices that restricts which memory each device can reach by DMA, used for security and device passthrough to VMs.
io_uring
A Linux asynchronous I/O interface using shared ring buffers between applications and the kernel to cut system-call overhead.
K
Kernel
The core of the operating system that runs in privileged mode and manages CPU, memory, devices and security.
Kernel module
Code, often a driver, that can be loaded into or removed from a running Linux kernel without rebooting.
kubelet
The agent on every Kubernetes node that starts Pods through the container runtime and reports their status.
KVM
Kernel-based Virtual Machine — the Linux kernel's built-in hypervisor, normally paired with QEMU.
L
Least privilege
Giving every user, process and service only the permissions it needs, for only as long as it needs them.
Live migration
Moving a running virtual machine from one host to another with little or no interruption.
Live patching
Applying critical kernel fixes to a running system in memory without an immediate reboot, for example kpatch or Canonical Livepatch.
M
Microkernel
An OS design that keeps only minimal services in the kernel and runs drivers and file systems as user-space servers.
Monolithic kernel
An OS design in which the scheduler, memory manager, file systems and drivers all run together in kernel space, as in Linux.
Mutex
A lock that lets only one thread at a time enter a critical section of code.
N
Namespace (Linux)
A kernel feature giving a process its own private view of a resource, such as process IDs, network interfaces, mounts or users.
O
OCI
Open Container Initiative — the Linux Foundation project publishing the image, runtime and distribution specifications for containers.
OOM killer
The Linux kernel mechanism that ends processes when memory runs out, either system-wide or inside a cgroup limit.
Operator
A Kubernetes controller plus custom resources that encode the operational knowledge for running a specific application.
Overlay file system
A union file system that stacks read-only image layers under a writable layer — the basis of container images.
P
Page cache
RAM the kernel uses to cache file data so repeated reads and writes avoid slow storage.
Page fault
A CPU exception raised when a program touches a virtual page that is not currently mapped, which the kernel then resolves or rejects.
Paging
Dividing memory into fixed-size pages (commonly 4 KiB) mapped through page tables from virtual to physical addresses.
Paravirtualization
Virtualization where the guest knows it is virtualized and uses efficient interfaces such as virtio instead of emulated hardware.
Pod
The smallest deployable unit in Kubernetes: one or more containers sharing a network namespace and storage volumes.
POSIX
The IEEE family of standards defining a common Unix-like system interface, shell and utilities.
Process
A running program with its own address space, open files, credentials and one or more threads.
Process control block (PCB)
The kernel data structure holding a process's state, registers, memory map, open files and scheduling information.
R
RBAC
Role-Based Access Control — granting permissions to roles and binding users or service accounts to them, as in Kubernetes.
Real-time OS (RTOS)
An operating system that guarantees tasks respond within strict, predictable time limits, such as FreeRTOS or Zephyr.
Reconciliation loop
A controller's continual cycle of comparing desired state with actual state and acting to close the gap.
S
Scheduler
The component that decides what runs next — which thread gets a CPU in an OS, or which node gets a Pod in Kubernetes.
SELinux
Security-Enhanced Linux — a mandatory access-control system that confines processes using labels and policy.
Semaphore
A synchronisation counter that limits how many threads can use a resource at once.
Service (Kubernetes)
A stable virtual IP and DNS name that load-balances traffic across the ready Pods matching a label selector.
Shell
A command-line interpreter such as bash, zsh or PowerShell that runs commands and scripts for the user.
SLI / SLO / SLA
Service level indicator (a measurement), objective (an internal target) and agreement (a contract with penalties).
Swap
Disk space (or compressed RAM, as with zram) used to hold memory pages that do not fit in physical RAM.
System call
The controlled entry point through which a program asks the kernel to do privileged work, such as read(), fork() or mmap().
systemd
The init system and service manager used by most Linux distributions, with units, targets, timers and the journal.
T
Thread
An independent flow of execution within a process, sharing its memory but with its own stack and registers.
TLB
Translation Lookaside Buffer — a CPU cache of recent virtual-to-physical address translations.
U
User space
The unprivileged CPU mode and memory region where applications run, separated from kernel space.
V
VFS
Virtual File System — the kernel layer giving one common interface (open, read, write) over many different file systems.
Virtual memory
Giving each process its own private address space, mapped by hardware and the kernel onto RAM and disk.
Z
Zombie process
A process that has exited but whose parent has not yet collected its exit status with wait().
Back to top