
Amazon S3
Scalable object storage service for unlimited data storage and retrieval with high durability and availability
Discover top open-source software, updated regularly with real-world adoption signals.

Scalable distributed storage system for modern infrastructure
Ceph delivers unified object, block, and file storage on commodity hardware. Designed for petabyte-scale deployments with self-healing and no single point of failure.

Ceph is a production-grade distributed storage platform that unifies object, block, and file storage into a single system. Built to run on commodity hardware, it scales from terabytes to exabytes while maintaining high availability through intelligent data replication and self-healing capabilities.
Ceph serves infrastructure teams, cloud providers, and enterprises requiring resilient storage at scale. Organizations running OpenStack, Kubernetes, or virtualization platforms rely on Ceph for persistent volumes, VM images, and S3-compatible object storage.
The platform eliminates single points of failure through its CRUSH algorithm, which distributes data across clusters without centralized metadata servers. Ceph's modular architecture includes RADOS (object store), RBD (block devices), CephFS (POSIX filesystem), and RADOS Gateway (S3/Swift API). Dual-licensed under LGPL 2.1/3.0, Ceph supports both development builds for testing and optimized production packages. Deployment options range from small test clusters using vstart to multi-datacenter installations managed via cephadm or orchestration tools.
When teams consider Ceph, these hosted platforms usually appear on the same shortlist.

Scalable object storage service for unlimited data storage and retrieval with high durability and availability

Massively scalable cloud object storage service for unstructured data (images, videos, backups) with high durability

Cloud file storage and sync for teams and individuals
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Kubernetes Persistent Storage
Dynamically provision RBD block volumes for stateful workloads with snapshot and clone capabilities
S3-Compatible Object Store
Replace proprietary cloud storage with on-premises RADOS Gateway supporting standard S3 APIs
OpenStack Block Storage
Serve Cinder volumes and Glance images with copy-on-write cloning for rapid VM provisioning
Shared Filesystem for HPC
Deploy CephFS for parallel workloads requiring POSIX semantics across compute clusters
Ceph runs on commodity x86_64 servers. Production clusters need at least 3 nodes for quorum. Each OSD daemon benefits from dedicated disks (HDDs or SSDs), and monitors require low-latency storage. RAM requirements scale with OSD count, typically 2-4 GB per OSD daemon.
Unlike centralized storage arrays, Ceph distributes data across cluster nodes with no single controller bottleneck. It scales capacity and performance by adding nodes, whereas traditional systems require forklift upgrades. Ceph trades operational complexity for horizontal scalability.
Most Ceph code is dual-licensed under LGPL 2.1 or 3.0. Some components use BSD-style licenses or are public domain. Documentation is CC-BY-SA-3.0. Check the COPYING file for per-file licensing details.
Yes. Rook is the recommended operator for running Ceph on Kubernetes, automating deployment and lifecycle management. Ceph also supports containerized deployments via cephadm using Podman or Docker.
Use RelWithDebInfo builds for production. Debug builds include extensive assertions and run at approximately 20% the performance of optimized builds. The do_cmake.sh script defaults to Debug when .git exists, so explicitly set CMAKE_BUILD_TYPE for production packages.
Project at a glance
ActiveLast synced 4 days ago