Summary
Ceph is an open-source, software-defined storage platform that delivers object, block, and file storage from a single unified cluster. It is designed to be self-healing, self-managing, and scalable to exabyte capacities.
What is Ceph?
Ceph stores data across a cluster of commodity servers using its CRUSH algorithm to distribute data without a central lookup table. This eliminates single points of failure and allows the cluster to grow by simply adding nodes.
The platform exposes three storage interfaces: RADOS Block Device (RBD) for block storage, CephFS for POSIX-compliant shared file systems, and RADOS Gateway (RGW) for an S3- and Swift-compatible object storage API. Kubernetes uses Ceph via the Rook operator to provide persistent storage for containerized workloads.
Ceph replicates or erasure-codes data across OSDs (Object Storage Daemons), automatically rebalancing and recovering when nodes are added or removed.
Why is Ceph relevant?
- Unified storage: One platform for block, file, and object storage reduces operational complexity
- Scalability: Scales horizontally from terabytes to exabytes without architecture changes
- Self-healing: Automatically detects and recovers from hardware failures
- Kubernetes integration: Widely used as the storage backend for Kubernetes clusters via Rook