Adoption & Ecosystem

Who uses it

The CNCF Incubator promotion blog (2021-11-04) names production users with described use cases (source 3).

Organisation	Use case	Source
Cerner	Healthcare IT; persistent storage and highly available data replication.	source 3
Tribunal Regional Eleitoral do Pará	Brazilian regional electoral court; storage backend for Prometheus and similar workloads.	source 3
Tyk	Open-source API and service management; backs hundreds of dynamically provisioned cluster nodes.	source 3

An honest counterpoint: not every adoption is a success story. Replicated removed Longhorn as the default storage in kURL, attributing drive corruption, mount failures, and unrecoverable state after reboot to Longhorn (source 11). Production guidance also stresses that Longhorn expects dedicated disks and replica-count tuning rather than defaults (source 12).

Adoption signals

Measured on 2026-06-24 via gh:

• longhorn/longhorn (umbrella): 7,805 stars, 712 forks, roughly 162 contributors. Community metrics concentrate on this umbrella repository.
• Individual implementation repos: longhorn-engine 386 stars, longhorn-manager 211, longhorn-instance-manager 27.
• The longhorn GitHub org holds 41 repositories.

For scale, the CNCF cited 34,000+ running nodes at promotion time (2021-11), and SUSE engineers later referenced 35,000 active nodes (source 3, source 10).

Ecosystem

• Kubernetes CSI: dynamic provisioning, snapshots, and volume expansion.
• RWX volumes: served through a built-in NFS share manager.
• Backups: to S3 or NFS targets; Velero integration for cluster backup and restore.
• Observability: Prometheus metrics.
• One-click install: from Rancher and SUSE Rancher Prime.
• Sibling data-plane repos: longhorn-engine (the v1 storage controller, "World's smallest storage controller", source 14) and longhorn-instance-manager (the per-node gRPC service that starts engine and replica processes), plus the SPDK-based v2 engine components.

Alternatives

The real distinction is scope. Longhorn is block-only and runs replica management itself; the alternatives trade that simplicity for breadth or raw speed (source 13).

Alternative	Differs by
Rook/Ceph	Serves block, file, and object from one system, at the cost of CRUSH map and placement-group learning curve and higher CPU overhead. Longhorn is block-only and simpler to operate, aimed at edge and mid-size clusters.
OpenEBS	Lets you pick engines: Mayastor (NVMe-oF/SPDK, fast) or cStor/Jiva (simpler). Longhorn ships one product and owns replica management end to end.
Portworx	Commercial; strong on application-aware snapshots and DR, but carries license cost.