To understand Ceph scalability and high availability, let's first talk about the architecture of traditional storage systems. Under this architecture, to store or retrieve data, clients talk to a centralized component known as a controller or gateway. These storage controllers act as a single point of contact for a client's request. The following diagram illustrates this situation:

This storage gateway, which acts as a single point of entry to storage systems, also becomes the single point of failure. This also imposes a limit on both performance and scalability while introducing a single point of failure, such that if the centralized component goes down, the whole system goes down too.

Ceph does not follow this traditional storage architecture; it has been totally reinvented for the next-generation of storage. Ceph eliminates the centralized gateway by enabling the clients to interact with the Ceph OSD daemons directly. The following diagram illustrates how clients connect to the Ceph cluster:

The Ceph OSD daemons create objects and their replicas on other Ceph Nodes to ensure data safety and high availability. Ceph also uses a cluster of monitors to ensure high availability, and to eliminate centralization, Ceph uses an algorithm called CRUSH, which stands for Controlled Replication Under Scalable Hashing. With the help of CRUSH, client on demand calculates where the data should be written to or read from. In the following recipe, we will examine the details of the Ceph CRUSH algorithm.