Harvester
harvesterhci.io
- Free plan
- Linux
- Web
- API
- Self-hosted
What Harvester is
Harvester is free, open-source hyperconverged infrastructure for bare-metal servers, combining virtualization with distributed storage. It targets virtual machines, Kubernetes running in VMs, containerized workloads on bare metal, and edge workloads. Its architecture uses Kubernetes, KubeVirt, Longhorn, Grafana, Prometheus, and SUSE Linux Micro. Users can create, edit, clone, and delete VMs, with SSH key injection, cloud-init, and graphical or serial consoles. Live migration can move a VM between hosts without downtime, but physical servers need matching CPU specifications. Distributed block storage and tiering are available; VM backups use NFS or S3-compatible destinations and cover Longhorn volumes only. Harvester installs directly onto bare metal from a bootable ISO, and new installations starting with v1.8 require UEFI boot. Three nodes are needed for multi-node capabilities and high availability. Production recommendations call for at least 16 CPU cores, 64 GB memory, 500 GB disk, and 10 Gbps Ethernet. Laptops and nested virtualization are unsupported.
Who it suits
Harvester suits teams running VMs, Kubernetes in VMs, containers on bare metal, or edge workloads on bare-metal servers. It is aimed at deployments able to meet its hardware and cluster requirements.
Good to know, and what to watch
- Free and open source, with no license fees stated.
- VMs support cloud-init and SSH key injection.
- Live migration moves VMs without downtime.
- Rancher can manage multiple Harvester clusters.
- Three nodes are required for high availability.
- Live migration requires matching physical CPU specifications.
- VM backups support Longhorn volumes only.
- Laptops and nested virtualization are unsupported.
Plans
Every plan and priceHarvester
Free plan
Getting started
- Download the Harvester bootable ISO
- Prepare supported ARM64 or x86_64 bare-metal servers
- Use UEFI boot for installations beginning with version 1.8
- Install Harvester directly onto the bare-metal servers
- Provide at least three nodes for high availability and multi-node features
- Open the web management interface and create virtual machines
Limits to check
A single-node cluster does not provide high availability, multiple replicas, or live migration. Full production recommendations call for at least 16 CPU cores, 64 GB memory, 500 GB disk, and 10 Gbps Ethernet; nested virtualization is unavailable.
The Club's review
Harvester is a free, open-source hyperconverged platform for bare-metal virtualization and distributed storage. It suits infrastructure teams running clustered VMs, Kubernetes and container workloads, including edge deployments. The trade-off is a server-oriented design whose hardware and cluster prerequisites exceed what small or laptop installations can provide.
Overview
Harvester combines virtual machines with distributed block storage in a Kubernetes-based stack built on KubeVirt and Longhorn, with Grafana and Prometheus for its operational architecture and SUSE Linux Micro at the core. That makes it a plausible consolidation platform for teams that want VM and container workloads managed together, rather than separate virtualization and storage products.
VM administration covers creating, editing, cloning and deleting machines, plus SSH-key injection, cloud-init, graphical and serial consoles, and snapshots. The immutable operating system protects the core image from direct edits outside designated configuration directories. This favors repeatable hosts and controlled change, but it also rules out ordinary in-place customization of the base system.
Key features
VM operations and migration
Live migration moves a running VM to another host without downtime, useful for maintenance and balancing work across a cluster. Physical servers must have matching CPU specifications, so heterogeneous hardware can constrain migration planning. Three nodes are needed to realize multi-node features and high availability; a single-node cluster has no high availability, multiple replicas or live migration.
Storage, backup and networking
Distributed block storage and tiering give VMs shared storage across the cluster. Backups target NFS or S3-compatible storage, but backup support is limited to Longhorn volumes, an important boundary for broader data-protection strategies. Networking supports virtual IPs, multiple NICs, VLANs and untagged networks. VLAN deployments require port trunking, so the physical network must be prepared accordingly.
Rancher and access control
Rancher can import and manage multiple Harvester clusters and place their VM administration alongside Kubernetes cluster management. Its authentication and role-based access controls suit multi-tenant administration, although Harvester's Rancher RBAC integration is documented as experimental, making it a feature to assess carefully for strict production governance.
Pricing
Harvester is free and open source. The Harvester plan costs free, with no license fees stated. That removes per-host software charges, but it does not remove the infrastructure cost of the recommended servers, networking and backup targets.
There are no paid tiers, seat quotas or trial periods to compare. The free plan is therefore the fit for organizations able to operate the platform themselves and accept its hardware and support-lifecycle responsibilities.
Platforms
Harvester runs self-hosted on bare-metal servers through a bootable ISO and provides a web management interface and API. Host architectures are ARM64 and x86_64. New installations beginning with version 1.8 require UEFI boot. Production recommendations call for at least 16 CPU cores, 64 GB of memory, a 500 GB disk and 10 Gbps Ethernet. Laptops and nested virtualization are unsupported. Guest support includes SUSE Linux Enterprise Server 15 SP6 and SP7 and SUSE Linux Micro 6.0 and 6.1.
Prime releases receive six months of active support followed by twelve months of maintenance support for critical security issues, which matters when planning upgrade and maintenance windows.
Who it's for
Harvester is aimed at infrastructure teams that need VMs, Kubernetes-in-VM deployments, containers on bare metal and edge workloads in one clustered system. It is particularly suitable where Rancher already coordinates several Kubernetes and virtualization environments, or where avoiding license fees is a priority.
It is a poor choice for a laptop, a nested lab, a single small server or a cluster assembled from mismatched CPUs. Those environments cannot use the platform's full availability, replication and migration capabilities, and may not meet the stated production hardware guidance.
Pros and cons
Pros
- One infrastructure layer: virtualization and distributed storage are combined around Kubernetes, KubeVirt and Longhorn.
- Broad VM tooling: cloning, cloud-init, SSH keys, consoles, snapshots and zero-downtime migration cover common operational tasks.
- No license fees: the free free plan lowers software acquisition cost for self-hosted teams.
- Multi-cluster administration: Rancher can manage Harvester clusters alongside Kubernetes clusters.
Cons
- Heavy minimum footprint: production guidance starts at 16 cores, 64 GB RAM, 500 GB disk and 10 Gbps networking.
- Cluster dependency: three nodes are required for high availability, replicas and live migration; single-node deployments lose those functions.
- Migration and backup limits: live migration needs matching CPUs, while VM backups cover Longhorn volumes only.
- Network and access caveats: VLANs require trunking, and Rancher RBAC integration is experimental.
Alternatives
For a wider comparison, see Hypervisor Software.
- Proxmox VE is the better alternative when you want a freemium virtualization platform spanning Linux, Windows, web and self-hosted environments, with a paid Premium plan.
- QEMU fits users seeking a free, GPL-licensed emulator and virtualizer across Linux, macOS and Windows rather than an integrated hyperconverged cluster.
- UTM is aimed at Apple users who want macOS or iOS virtualization, with a free open-source build or a $9.99 one-time Mac App Store purchase.
- crosvm is a free BSD-3-Clause open-source option for Android, Linux, Windows, self-hosted and API deployments.
- KVM is the choice for a Linux-native open-source hypervisor integrated with the mainline Linux kernel and QEMU userspace.
- Multipass suits lightweight local cloud launches across Linux, macOS and Windows.
- oVirt provides a free community virtualization edition for API, Linux, self-hosted and web environments.
- VMware Workstation Pro is a free option for 64-bit Intel or AMD Windows and Linux hosts without a subscription or license key.
Verdict
Choose Harvester when you need clustered bare-metal VMs and distributed storage, especially alongside Rancher-managed Kubernetes, and can meet its three-node and server-grade hardware requirements. Look elsewhere if you need a lightweight single-host hypervisor, mixed-CPU live migration, broader backup coverage or laptop and nested-virtualization support.
The verdict
Harvester combines VM management and distributed storage on bare-metal servers, with additional cluster management through Rancher. Plan for the node count, hardware recommendations, and storage limits before choosing it.
Questions people ask
Does Harvester cost anything?
Harvester is free and open source. Its plan is listed as free, with no license fees stated.
What does it run on?
Harvester installs on bare-metal servers and supports ARM64 and x86_64 host platforms. Laptops and nested virtualization are unsupported.
How many nodes are required?
Three nodes are required to fully realize multi-node features and high availability. Single-node clusters do not support multiple replicas or live migration.
Can Harvester migrate running VMs?
Yes. Live migration moves a VM to another host without downtime, but the physical servers need matching CPU specifications.
Where can VM backups be stored?
Backups use NFS or S3-compatible targets, and backup support is limited to Longhorn volumes.
Does it integrate with Rancher?
Rancher can import and manage multiple Harvester clusters and their VMs alongside Kubernetes clusters. Rancher role-based access integration for Harvester is documented as experimental.
What the maker publishes
- Cluster requirements
- Three nodes are required to fully realize multi-node features and high availability; single-node clusters do not support high availability, multiple replicas, or live migration. source
- Deployment
- Harvester installs directly on bare-metal servers from a bootable ISO; new installations starting with v1.8 require UEFI boot. source
- Hardware limits
- The production recommendations include at least 16 CPU cores, 64 GB memory, 500 GB disk, and 10 Gbps Ethernet; laptops and nested virtualization are unsupported. source
- Intended workloads
- Harvester is aimed at running VMs, Kubernetes in VMs, containerized workloads on bare metal, and workloads at the edge. source
- Live migration
- Harvester supports moving a VM to another host or node with zero downtime, and live migration requires matching CPU specifications across physical servers. source
- Networking
- Harvester supports virtual IPs, multiple network interface cards, VLAN or untagged networks, and requires port trunking for VLAN support. source
- Operating system
- Harvester uses an immutable operating system that prevents direct changes to the core system image outside designated configuration directories. source
- Product
- Harvester is open-source hyperconverged infrastructure for bare-metal servers, combining virtualization and distributed storage. source
- Rancher integration
- Rancher can import and manage multiple Harvester clusters and manage their VMs alongside Kubernetes clusters. source
- Security and access
- Rancher integration supports authentication and role-based access control for multi-tenancy; Harvester's Rancher RBAC integration is documented as experimental. source
- Storage and backup
- Harvester provides distributed block storage and tiering; VM backups use NFS or S3-compatible targets, and backup support is limited to Longhorn volumes. source
- Support lifecycle
- The maker's docs state that Prime releases receive six months of active support followed by twelve months of maintenance support for critical security-related issues. source
- Technology
- Its architecture uses Kubernetes, KubeVirt, Longhorn, Grafana, and Prometheus, with SUSE Linux Micro at its core. source
- VM management
- The platform supports creating, editing, cloning, and deleting VMs, with SSH key injection, cloud-init, and graphical and serial consoles. source
Checked against the maker's pages on 2 Oct 2026.