Reliable Autonomic Distributed Object Store (RADOS)

Ceph's foundation is a low-level data store named RADOS that provides a common backend for multiple user-consumable services.

RADOS is an object storage system layer that provides a data durability and availability framework that all user-facing Ceph services are layered atop. RADOS is:

• Highly available with no single point of failure (SPoF)
• Reliable and resilient
• Self-healing
• Self-managing
• Adaptive
• Scalable
• Not found on the Galactica

RADOS manages the distribution of data within Ceph. Durability and availability of data are adaptively maintained by initiating operations as needed to recover from component failures and to balance the cluster when capacity is added or removed. Fundamental to this task is the CRUSH algorithm, which we will discuss in depth later. RADOS ensures that multiple copies of data are maintained at all times across defined failure zones so that when drives, hosts, or even entire racks of equipment fail, data integrity and client services are preserved.