All Tools
L
DataFreeOpen Source
LONGHORN
Kubernetes-native distributed block storage
Apache-2.0
ABOUT
GPU jobs, vector databases, and model caches need persistent disks on Kubernetes, but cloud volumes are expensive and a SAN is operationally heavy. Longhorn runs replicated block storage as pods on the cluster itself, with snapshots, backups to S3, and a UI, so PVs survive node failure without a separate storage appliance.
INTEGRATION GUIDE
1. Give vector databases and feature stores replicated PersistentVolumes on Kubernetes
2. Snapshot and back up model-cache disks to S3 from a Longhorn volume
3. Run GPU training jobs that need durable local-like block storage across nodes
4. Recover inference stateful sets after a node dies without a SAN
TAGS
kubernetesblock-storagepersistent-volumesstoragecncfmlopsopen-source