When you are building up a storage infrastructure, scalability is one of the most important design aspects. The storage solution that you have chosen for your infrastructure should be scalable enough to accommodate your future data needs. Usually, a storage system starts with small to medium capacity and grows gradually into a large storage solution.

Traditional storage systems were based on scale-up design and were limited by a certain storage capacity. If you try to expand these storage systems over a certain limit, you might need to compromise the performance, reliability, and availability. The scale-up design methodology for storage involves adding disk resources to the existing controller systems, which becomes a bottleneck for performance, capacity, and manageability when it reaches a certain level.

On the other hand, scale-out design focuses on adding entire new devices containing disks, CPU, memory, and other resources to the existing storage cluster. With this type of design, you don't face the challenges that have been seen in scale-up design; it rather benefits from linear performance improvement. The following diagram explains the scale-up and scale-out design of a storage system:

Ceph is a seamless scalable storage system based on scale-out design, where you can add a compute node with a bunch of disks to an existing Ceph cluster and extend your storage system to a larger storage capacity.