Canonical MAAS
canonical.com
- Free plan
- Open source
- Linux
- Web
- API
- Self-hosted
What Canonical MAAS is
Canonical MAAS is an open-source platform for turning physical servers into centrally managed, cloud-like resources. It handles the server lifecycle, from discovering and inventorying hardware to commissioning, deployment, and decommissioning. Teams can manage nodes through a web interface, command line, or API, and connect infrastructure workflows with Terraform, Juju, and Ansible. Hardware discovery uses PXE boot, supports IPMI-enabled machines, and can cover subnets, VLANs, and switch fabrics. MAAS configures networking and storage, runs hardware tests, and manages LXD and KVM virtual machines. Its listed images include Ubuntu, CentOS, Windows, and RHEL, with synchronization or onsite mirrors available for offline use. MAAS is suited to private cloud with OpenStack, server automation, and bare-metal Kubernetes. A free self-support plan is available; paid plans start at $30 a year per managed machine. Ubuntu Pro is required for machines hosting MAAS controllers.
Who it suits
MAAS suits teams building private clouds with OpenStack, automating server operations, or deploying Kubernetes on bare metal. Its single-machine installation option also fits homelabs and proofs of concept.
Good to know, and what to watch
- Automates discovery through decommissioning of physical servers.
- Offers web, command-line, and API management.
- Integrates with Terraform, Juju, and Ansible.
- Supports network and storage configuration.
- Can manage LXD and KVM virtual machines.
- Controllers require Ubuntu Pro.
- Free self-support has no HA support or RBAC.
- Free plan includes no phone or ticket support.
Plans
Every plan and priceFree
Self-support
Essential
Standard
Free plan
The Club's review
Canonical MAAS is an open-source control plane that turns physical servers into centrally managed, cloud-like resources. It is aimed at infrastructure teams operating bare-metal fleets, private clouds, or Kubernetes platforms, especially where repeatable provisioning matters. The broad automation is persuasive, but controller support requirements and per-machine pricing shape its fit.
Overview
MAAS covers the server lifecycle from initial discovery and inventory through commissioning, deployment, and eventual decommissioning. That makes it more than an imaging utility: it provides a management layer for hardware that would otherwise be handled through separate tools and procedures. Canonical positions it for private cloud with OpenStack, server automation, and bare-metal Kubernetes.
Installation can be a single region-and-rack machine for a homelab, proof of concept, or small production site. Larger environments can use multiple controllers for high availability, although HA support is tied to paid plans. MAAS is self-hosted and available through a web interface, command line, and API, with Linux-based deployment and network provisioning.
Key features
PXE boot discovers servers, while IPMI support helps bring remote hardware into management. Discovery extends across subnets, VLANs, and switch fabrics, useful when equipment spans more than one network segment. Administrators can operate nodes interactively in the web UI or automate them through the CLI and API.
The cloud-style API connects with Terraform, Juju, and Ansible, so bare-metal changes can be incorporated into infrastructure-as-code workflows. Ubuntu, CentOS, Windows, and RHEL images can be synchronized or served from an onsite mirror, an important option for offline or restricted sites.
Networking includes address assignment, bonding, VLANs, virtual bridges, static routes, DHCP, and DNS management. Storage layouts cover Bcache, RAID, and LVM. LXD and KVM integration adds virtual-machine creation and management through the same UI, CLI, and API, although organizations seeking only VM administration may find MAAS broader than necessary.
Built-in or uploaded hardware tests can run during operations, with historical results and metrics retained for later review. That gives teams an evidence trail for unreliable equipment rather than relying solely on deployment failures to reveal problems. MAAS also supports unattended deployment, driver injection, multi-OS provisioning, and multi-tenant management.
Ubuntu Pro supplies security patching and enterprise support for MAAS controllers and Ubuntu machines, and controllers require Ubuntu Pro. Standalone Support can cover machines without Ubuntu Pro, but the machines hosting the controllers still need it.
Pricing
| Plan | Price | What it means |
|---|---|---|
| Self-support | Free | Free for all machines; no HA, RBAC, Knowledge Base, Sev 1 response SLA, phone support, or ticket support. |
| Essential | $30 a year | Billed per managed machine; adds HA, RBAC, and Knowledge Base access, but no Sev 1 response SLA, phone support, or ticket support. |
| Standard | $50 a year | Billed per managed machine; includes HA, RBAC, Knowledge Base, a four-hour Sev 1 response time, and phone and ticket support during office hours. |
| Advanced | $100 a year | Billed per managed machine; includes the same core capabilities, a one-hour Sev 1 response time, and phone and ticket support 24 hours a day, every day. |
The free tier is practical for labs or teams able to operate without resilience and support entitlements. Essential is the entry point for production HA and role-based access, but organizations needing direct assistance must move to Standard. Advanced is aimed at operations where round-the-clock support and a one-hour Sev 1 target justify the highest per-machine charge.
Platforms
MAAS is self-hosted, Linux-based, web-accessible, and API-driven. It provisions physical hardware over the network and supports both server types and operating-system families, including bare-metal deployment and virtual-machine management.
Who it's for
Choose MAAS for a private cloud, automated server estate, or bare-metal Kubernetes environment where hardware lifecycle control and repeatability are central requirements. It also suits teams combining interactive administration with Terraform, Juju, or Ansible automation. A homelab or proof of concept can start with one machine; production sites needing controller resilience should budget for a paid plan and Ubuntu Pro on controller hosts.
Pros and cons
Pros
- End-to-end physical-server lifecycle automation reduces separate discovery, imaging, and retirement processes.
- PXE, IPMI, and fabric-aware discovery accommodate heterogeneous infrastructure.
- Web, CLI, and API access supports both operators and repeatable automation.
- Broad networking, storage, mixed-OS images, VM integration, and hardware-test history cover complex environments.
- The free plan allows all machines to be managed without a license fee.
Cons
- The free tier omits HA, RBAC, knowledge resources, formal Sev 1 response, phone support, and ticket support.
- HA and support are charged per managed machine, so larger fleets can become expensive.
- MAAS controllers require Ubuntu Pro, adding an operational subscription requirement.
- Teams wanting only simple imaging may not need its wider infrastructure scope.
Alternatives
For a broader shortlist, see OS Deployment Software, Bare Metal Provisioning Software, and IT Automation Software.
- OPSI is worth considering when a free community tier, cross-platform clients, or a 30-device basic plan better matches the rollout.
- iVentoy fits straightforward PXE deployment, especially where a free edition with a 20-client cap and no commercial use is acceptable.
- Acronis Cyber Disaster Recovery Cloud is the alternative when disaster-recovery cloud storage and compute, billed pay-as-you-go, are the priority.
- AOMEI Backupper is a Windows-focused backup choice with a perpetually free Standard edition.
- Clonezilla is a free, GPL-licensed option for imaging rather than a full MAAS-style management plane.
- FOG Project offers free, open-source deployment with community support.
- Foreman is a free, self-managed open-source alternative for teams that prefer its provisioning approach.
- Serva suits Windows-based, smaller-scale PXE needs, with its free community plan limited to personal, non-commercial use and two clients.
Verdict
Canonical MAAS is a strong choice for infrastructure teams that need programmable, multi-OS bare-metal provisioning tied to hardware discovery, network and storage configuration, testing, and lifecycle control. Its main reason to look elsewhere is cost and operational commitment: production HA and support require per-machine subscriptions, while controllers require Ubuntu Pro.
The verdict
MAAS brings server provisioning and ongoing hardware management into one platform, with automation and multiple management interfaces. Check Ubuntu Pro requirements and support-plan limits when choosing an operating model.
What the maker publishes
- API
- MAAS exposes a REST API with GET, POST, PUT and DELETE methods; version 2.0 is the documented supported API version. source
- API authentication
- API authentication uses 0-legged OAuth, and the documentation provides Python, Ruby, cURL and HTTPie examples. source
- API coverage
- The latest reference states that every feature available in the UI is also accessible through the API and that the v2 reference is generated from OpenAPI. source
- Architecture
- A PostgreSQL-backed region controller handles operator requests while distributed rack controllers provide DHCP, IPMI, PXE, TFTP and cached images. source
- Company founded
- Canonical's LinkedIn profile lists 2004 as its founding year. source
- Company headquarters
- Canonical's LinkedIn profile lists London, England as headquarters. source
- Company legal name
- Crunchbase lists the maker's legal name as Canonical Ltd. source
- Company ownership
- Crunchbase describes Canonical as a privately held, for-profit company and identifies Mark Shuttleworth as founder and CEO. source
- Company size
- LinkedIn places Canonical in the 1,001–5,000-employee range. source
- Device discovery
- MAAS discovers devices and interfaces passively, on demand or periodically. source
- Documentation depth
- The documentation is organized into tutorials, how-to guides, explanations and reference material, including installation, API, Terraform, troubleshooting, metrics, logging and security-hardening sections. source
- Extensibility
- Canonical documents CLI and Python client bindings, custom commissioning scripts, deployment scripts and custom power-driver plugins. source
- Founded
- 2004 source
- Hardware discovery
- MAAS discovers servers through PXE boot and supports IPMI-enabled machines, with discovery for subnets, VLANs, and switch fabrics. source
- Hardware discovery
- PXE boot can automatically discover and enlist servers, while network discovery identifies subnets, VLANs and switch fabrics. source
- Hardware inventory
- Commissioning records processors, memory, storage, networking and other components, with lifecycle event tracking. source
- Hardware testing
- MAAS can run built-in hardware tests or uploaded tests and retain historical test data and metrics. source
- Hardware testing
- Administrators can run supplied hardware tests or upload their own and review historical component metrics. source
- Headquarters
- London, United Kingdom source
- High availability
- Region-controller high availability is database-based, and rack controllers can provide local service redundancy. source
- Infrastructure as code
- MAAS provides a cloud-like API and integrates with Terraform, Juju, and Ansible for infrastructure automation and orchestration. source
- Installation options
- Canonical documentation describes a single-machine region-and-rack installation for homelabs, proofs of concept, or small production environments, and multiple controllers for high availability. source
- Integrations
- Canonical documentation names Juju, Chef, Puppet, Ansible, Terraform, OpenStack and Kubernetes as MAAS automation or orchestration integrations. source
- License
- The MAAS source repository is distributed under the GNU Affero General Public License version 3. source
Checked against the maker's pages on 15 Oct 2026.