Harvester
Harvester is an open source Hyperconverged Infrastructure (HCI) platform (infrastructure virtualization) built on Kubernetes, designed to run and manage Virtual Machine (VM) workloads with cloud-native tooling on bare metal servers.
- HCI platform for virtual machines on bare metal (infrastructure virtualization).
- Kubernetes-based control plane for scheduling and lifecycle of virtual machines (container orchestration / VM orchestration).
- Integrated block storage using Longhorn for highly available VM volumes (storage orchestration).
- Support for Rancher integration for multi-cluster management and self-service provisioning (cloud management / platform operations).
- Network management for VM workloads, including virtual networks and load balancing capabilities (software-defined networking).
More About Harvester
Harvester addresses the problem space of consolidating compute, storage, and networking for VM workloads into a single software-defined stack (hyperconverged infrastructure) while using a Kubernetes-based architecture familiar to cloud-native operations teams.
The platform runs directly on bare metal servers and provides virtualization (infrastructure virtualization) so that users can provision and operate virtual machines without a separate proprietary hypervisor layer. It uses Kubernetes (container orchestration) as its core orchestration and management framework, which allows Harvester to expose VM lifecycle operations, scheduling, and high availability using Kubernetes primitives and controllers tailored for virtual machines.
For storage, Harvester integrates Longhorn (storage orchestration), a CNCF project from SUSE, to provide distributed block storage for VM volumes. Longhorn provides replicated volumes across nodes and supports operations such as snapshots and backups, which Harvester uses to support VM resilience and data protection. This integration allows Harvester clusters to use commodity local disks on each node to form a unified storage pool for VM workloads.
Networking in Harvester (software-defined networking) includes virtual networks and support for multiple network interfaces for virtual machines, enabling separation of traffic types and integration with existing data center networks. Load balancing and ingress patterns are supported through the Kubernetes ecosystem, allowing exposure of VM-based services in a manner similar to container workloads.
Harvester integrates with Rancher (cloud management / platform operations), SUSE’s Kubernetes management platform, to provide centralized cluster management, access control, and multi-tenant operations. Through Rancher, platform teams can manage Harvester clusters alongside other Kubernetes clusters, standardizing governance, authentication, and observability across environments.
Enterprises use Harvester in data center and edge environments to run traditional VM workloads, sometimes in parallel with container workloads, while relying on a cloud-native operating model. The platform supports VM templates, images, and ISO management for Operating System (OS) deployment (image management), as well as features such as live migration and high-availability policies where documented (availability management).
From a directory and taxonomy perspective, Harvester fits into categories including HCI platforms, Kubernetes-based virtualization, and software-defined storage and networking for virtualized workloads. It is relevant for organizations that want a unified stack for VM operations on bare metal, with integration into Kubernetes and Rancher-based cluster management.