Longhorn Storage Solution¶
DCE 5.0 supports many third-party storage solutions. We have conducted detailed tests on Longhorn and finally integrated it into the app store as an Addon. The following is a brief research and evaluation report on Longhorn.
Longhorn is a lightweight cloud native Kubernetes distributed storage platform that can run on any infrastructure. Longhorn and DCE can be used together to deploy high-availability persistent block storage.
-
Design and Architecture
- Control plane: Longhorn Manager deployed with DaemonSet
- Data plane: Longhorn Engine is a storage controller that can have multiple replicas
-
Longhorn Storage Volumes
- Support for Thin provisioning of storage volumes
- Storage volume maintenance mode is used for snapshot reverting operation
- Each volume replica contains multiple snapshots.
- The default number of replicas of the storage volume can be set in settings. The number of replicas can be changed through the UI after the storage volume is mounted.
- Longhorn is a crash-consistent block storage solution that automatically synchronizes the sync command before creating a snapshot snapshot
-
Data backup and external secondary storage
- NFS/S3 compatible external secondary storage for backups is independent of the Kubernetes cluster. Data remains available even if the Kubernetes cluster is unavailable
- Longhorn will also synchronize the storage volume to the secondary storage of the disaster recovery cluster (DR) for disaster data recovery
- A backup is a flattened collection of multiple snapshot data.
- Supports continuous and repeated snapshots and backups.
- Support Clone of CSI storage volume
-
High Availability
- Support Replica automatic balance setting
- Support data locality setting: there is at least one replica copy on the node running the pod using the storage volume
- Support for displaying node storage space usage.
- Support Kubernetes Cluster Autoscaler (Experimental)
- Supports automatic recovery after storage volumes are accidentally unmounted
- Supports automatic recovery of storage volumes after cluster node failure
-
Monitoring Monitoring
- Support Prometheus and Grafana to monitor Longhorn
- Longhorn metrics can be integrated into DCE monitoring system
- Support Kubelete Metrics monitoring
- Support Longhorn alert strategy
-
Advanced features
- Support Backing Image
- Support for Orphaned Replica Directories
- Support DCE cluster recovery: recovery of all storage volumes in the cluster
- Support multiple write operations ReadWriteMany (RWX) workloads (NFSv4)
- Support Longhorn Volume as iSCSI Target