OpenStack

What it is

OpenStack is an open-source cloud infrastructure platform for provisioning and managing compute, storage, and networking resources — effectively a DIY alternative to AWS/Azure/GCP for organizations that want to run their own IaaS. In the Kubernetes ecosystem, it shows up most often as the substrate underneath clusters: Cluster API's OpenStack provider, Magnum (OpenStack's own k8s-as-a-service project), and various CSI/CCM plugins that let K8s talk to Cinder, Neutron, and Nova. It's not a Kubernetes-native tool — it predates Kubernetes by years — but it's frequently the layer Kubernetes gets provisioned on top of in private cloud and telco environments.

A canonical description isn't publicly available from the project in a single-line form because OpenStack is a federation of ~40+ independent services (Nova, Neutron, Cinder, Keystone, Glance, etc.) rather than one monolithic product.

Who builds it and why

OpenStack is governed by the Open Infrastructure Foundation (OIF), with contributions historically dominated by large telcos (AT&T, China Mobile, Deutsche Telekom), infrastructure vendors (Red Hat, Canonical, Mirantis, Huawei), and public sector/research institutions running private clouds at scale. The motivation is consistent across that base: control over infrastructure, data sovereignty, and avoiding hyperscaler lock-in — not developer experience or speed of iteration.

4,560 contributors is a genuinely large number for any project, reflecting OpenStack's age and the breadth of sub-projects under its umbrella rather than a single tight-knit maintainer team. Expect contribution patterns to be uneven across components — Nova and Neutron get heavy attention; smaller services vary.

Production readiness signal

• 5,962 GitHub stars is low relative to OpenStack's actual install base and real-world usage — this is an artifact of OpenStack using its own Gerrit-based review system historically, with GitHub as a mirror, not the primary development venue. Don't read star count as a popularity signal here.
• Last commit 2026-07-01 indicates active, ongoing development.
• 4,560 contributors signals broad organizational investment, consistent with a project that has run production telco and enterprise clouds for over a decade.
• CNCF maturity: not applicable — OpenStack is not a CNCF project (it sits under the Open Infrastructure Foundation).
• Latest release, founding date: not publicly available in the data provided — check releases.openstack.org and the project's own history directly before making a deployment decision.

Bottom line: this is a mature, battle-tested, operationally heavy project. The activity signals are real; the GitHub metrics undersell it.

Who should use this

• Telcos and service providers building NFV/edge infrastructure who need bare-metal-adjacent control over compute and networking.
• Enterprises and government entities with hard data residency or air-gapped requirements that rule out public cloud.
• Organizations already running OpenStack that need to provision Kubernetes on top of it (via Magnum or Cluster API) rather than migrate off it.
• Teams with dedicated infrastructure/ops staff who can own upgrades, networking (Neutron is notoriously complex), and the operational burden of running 15-40 interacting services.

Who should NOT use this

• Small-to-mid teams without dedicated platform/infra engineers — OpenStack's operational complexity (Neutron networking, Keystone auth, service interdependencies) will consume more engineering time than it saves versus public cloud or a lighter virtualization layer.
• Anyone whose primary goal is running Kubernetes, full stop — if you don't already have OpenStack or a mandate requiring it, provision K8s directly on bare metal, a hyperscaler, or a simpler virtualization stack (Proxmox, bare-metal + Cluster API).
• Greenfield projects prioritizing speed — OpenStack's setup and upgrade cycles are measured in weeks/months, not hours.
• Cost-sensitive small deployments — the hardware and staffing overhead rarely pays off below a certain scale (roughly hundreds of nodes).

Alternatives

• Bare-metal + Cluster API / Metal3 — provision Kubernetes directly without an OpenStack layer; less operational overhead if K8s is the actual goal.
• VMware vSphere / Proxmox — simpler virtualization layers for private infrastructure when you don't need full multi-tenant IaaS.
• Public cloud (AWS/Azure/GCP) — if data sovereignty and control aren't hard requirements, hyperscalers eliminate the operational burden entirely.

Pricing

Fully open source under Apache-2.0 — no licensing cost. Real cost is operational: hardware, staffing, and the engineering time to run and upgrade the platform. Commercial distributions (Red Hat OpenStack Platform, Canonical, Mirantis) exist if you want vendor support instead of running it raw.