Proxmox vs Harvester
Proxmox VE and SUSE Harvester compared on architecture, hardware needs, storage, Kubernetes and day-to-day operations, and which one suits your team.
Updated
The short answer
Choose Proxmox VE for traditional virtualization that runs on modest hardware and is easy for Linux administrators to learn. Choose Harvester if your organization already works in Kubernetes and wants VMs managed the same way.
Both are open source and both run virtual machines on KVM. The difference is what sits underneath: Proxmox VE is a Debian-based virtualization manager, while Harvester, which SUSE also sells as SUSE Virtualization, is built on Kubernetes.
Side by side
| Proxmox VE | Harvester | |
|---|---|---|
| Foundation | Debian, with KVM and LXC | Kubernetes on bare metal, with VMs through KubeVirt |
| Storage | ZFS, LVM, Ceph, NFS and more | Longhorn distributed block storage |
| High availability | At least three votes: three nodes, or two plus a QDevice | A three-node cluster; a single node has no HA, replicas or live migration |
| Production hardware | Ordinary servers; needs depend on your workloads and storage | Per node: at least 16 cores, 64 GB of memory, 500 GB of disk and 10 Gbps networking |
| Containers | LXC system containers | Kubernetes workloads, with guest clusters through Rancher |
| Skills needed | Linux administration | Linux, and Kubernetes for anything beyond the basics |
| Licence | AGPLv3 | Apache 2.0 |
| Paid support | Proxmox and its partners, per CPU socket | SUSE |
What running each one is like
Proxmox VE installs on each node and clusters with a few steps. It has a long track record, a large user base and a simple model: nodes, storage and VMs. It fits small clusters well, including ones with limited hardware.
Harvester asks for more hardware per node, partly because storage replication and Kubernetes itself use resources. In return, VMs become Kubernetes objects, which suits teams that already manage everything through Kubernetes tooling. When something breaks, though, you are debugging Kubernetes, KubeVirt and Longhorn together, so that experience needs to exist on the team.
Choose Proxmox VE if
- Your team knows Linux and virtualization, and not necessarily Kubernetes.
- Your hardware is modest or mixed.
- You want a choice of storage, including ZFS on a single node or Ceph across several.
- You want LXC containers alongside VMs.
Choose Harvester if
- Your organization already runs Kubernetes and Rancher.
- You want VMs and clusters managed through the same tools.
- You can provide at least three well-specified nodes and fast networking.
If you want self-service without building it
Both products leave you to build the cloud experience on top. CloudOps gives servers you own self-service VMs, Kubernetes clusters, private networks, load balancers and S3-compatible storage, from a console, an API and Terraform, and we run the control plane as a service.
It does less in places. CloudOps has no live migration between hosts and no automatic VM restart on another host, it cannot import Proxmox VMs, and you cannot install the control plane yourself.
More detail: CloudOps as a Proxmox alternative and Proxmox vs OpenStack.