Zero is in development. Everything on this page is planned.

Concepts

Zero uses the name the industry already uses for the same concept. Each term below gives Zero’s meaning, and the name it matches elsewhere.

Zero OS runs on each machine. Zero Platform runs your workloads across them.

Machines

node

A machine, physical or virtual, running Zero OS.

Like: Kubernetes node, Nomad client node, Proxmox node

cluster

Nodes managed together by one Zero Platform.

Like: Kubernetes cluster, Proxmox cluster

Running workloads

workload

Something you deploy, described by its spec.

Like: Kubernetes workload resource, Nomad job

pod

One running copy of a workload, on one node. Its containers start, stop and move together and share one network address.

Like: Kubernetes pod, Podman pod, Nomad allocation

container

One isolated process tree in a pod, with its own root, user, cgroup and sandbox.

Like: Linux or OCI container

sandbox

What Zero OS runs a pod in: the pod’s own kernel, created and confined by Zero OS. A fence of namespaces, cgroup, Landlock, seccomp and its own user surrounds each container inside.

Like: Kubernetes pod sandbox (CRI), Kata sandbox, Firecracker microVM

Desired state

spec

The plain-data description of a workload.

Like: Kubernetes spec

revision

An immutable, kept version of what a workload’s pods run. Replicas, spread, rollout strategy and schedule change without one.

Like: ECS task definition revision, Kubernetes pod template in a ReplicaSet

namespace

A named scope in which workload names are unique. The same name can exist in several namespaces. Other workloads reach a workload by its service name:

<workload>.<namespace>

Like: Kubernetes namespace

desired state, current state

What was asked for, and what actually runs.

Like: Kubernetes, Borg