longhorn/longhorn
Longhorn
Longhorn is a distributed block storage system and orchestrator for Kubernetes. It provides persistent, replicated block storage volumes that survive pod restarts and node failures by maintaining synchronous copies of data across multiple cluster nodes.
The system implements the Container Storage Interface (CSI) for dynamic volume provisioning and attachment. It is distinguished by its support for shared read-write access to a single block volume across multiple pods, as well as the ability to export volume snapshots to external S3 or NFS targets for off-cluster disaster recovery.
The platform covers a broad set of storage management capabilities, including point-in-time snapshotting, instant volume cloning, and automated backup scheduling. It includes tools for node-level disk pooling, volume group management, and health monitoring of capacity and performance through a dedicated dashboard.
Deployment and management are handled through a Kubernetes-native control plane using custom resource definitions or via a curated application marketplace.
Features
- Kubernetes Storage Orchestration - Provides a distributed block storage system and orchestrator for managing persistent volumes in Kubernetes.
- Block Storage - Provides a distributed block storage system for persistent volumes across Kubernetes clusters.
- Point-In-Time Snapshots - Captures point-in-time, immutable views of volumes for rapid recovery and consistent reads.
- CSI Driver Implementations - Acts as a Kubernetes CSI driver to enable dynamic provisioning and attachment of block volumes.
- Synchronous Replication Management - Ensures immediate consistency and fault tolerance by requiring confirmation from multiple replicas before committing writes.
- Replicated Block Volumes - Maintains synchronously replicated block volumes across multiple nodes to ensure high availability.
- Distributed Replica Orchestration - Manages volume replicas as independent engine processes that synchronously replicate writes across nodes.
- Persistent Storage Volumes - Connects block storage volumes to specific pods to make data available for processing.
- Replicated Persistent Volumes - Creates persistent volumes that synchronously replicate block data across multiple nodes for high availability.
- Storage Volume Orchestration - Implements automated provisioning and mounting of persistent block storage to containerized workloads.
- CRD-Based Resource Management - Provides a Kubernetes-native control plane that manages storage volumes and replicas using custom resource definitions.
- CSI Driver Implementations - Exposes block storage volumes to Kubernetes through the Container Storage Interface standard for dynamic provisioning.
- Kubernetes Controllers - Uses Kubernetes controllers and CRDs to manage the distributed state of storage volumes and replicas.
- Storage Replica Recovery - Automatically reattaches and rebuilds volume replicas when a cluster node becomes unavailable.
- Object Storage Backups - Provides the ability to export volume snapshots to external S3 or NFS targets for off-cluster disaster recovery.
- Volume Snapshot and Clone Engines - Implements an engine for capturing point-in-time volume states and creating independent clones.
- Snapshot Rollbacks - Rolls back a volume to a previous point-in-time state using a stored snapshot.
- Disaster Recovery Workflows - Provides automated workflows for exporting volume backups to external S3 or NFS targets for disaster recovery.
- Differencing Snapshot Chains - Stores volume snapshots as a chain of read-only differencing layers for instant rollback and efficient backups.
- Access Mode Management - Coordinates shared read-write access for a single block volume across multiple pods simultaneously.
- Object Storage Volume Exports - Serializes volume snapshots to external object storage targets for off-cluster disaster recovery.
- Management Dashboards - Provides a dedicated dashboard to create, attach, snapshot, backup, and delete persistent volumes.
- Image-Based Provisioning - Provisions persistent volumes by extracting block data from container images for rapid deployment.
- Read-Only Access Modes - Allows multiple nodes to read from a single volume concurrently while preventing write operations.
- Read-Write-Many Block Storage - Provides shared read-write access to a single block volume across multiple pods simultaneously.
- Placement Optimizations - Optimizes the placement of data replicas across nodes based on scheduling strategies for efficiency.
- Local Disk Aggregation - Aggregates physical disks on individual cluster nodes into storage pools for replica placement.
- Backup Job Scheduling - Schedules recurring backup jobs to copy volume snapshots to external storage targets at defined intervals.
- Zero-Downtime Engine Upgrades - Allows transitioning volumes to newer data engine versions without requiring application downtime.
- Storage Pool Health Monitors - Tracks volume capacity, IOPS, and replication status through specialized storage pool health monitors.
- Node Storage Pooling - Manages physical disks on each Kubernetes node as storage pools for placing volume replicas.
- Storage - Distributed block storage system for Kubernetes.
- Storage Management - Distributed block storage system for Kubernetes.
