Software-Defined Storage with Ceph

From scientific data to food pictures, we're creating more data than ever before, and we need a place to store it all. While one can easily spend big bucks with a proprietary storage vendor, there are free and open source solutions available, the most versatile being Ceph.

Ceph is a fault-tolerant, scale-out, software-defined storage solution designed to run on commodity servers and disk drives. Ceph provides object, block, and filesystem storage, suitable for cloud service data, VMs, or regular files in a way that's resilient against drive or server failures.

Based on desired workload, Ceph clusters can be designed for varyting degrees of speed, density, and resiliancy, in different proportions, even on the same cluster. With data organized into pools, different pools can be tuned for different characteristics with, for example, some pools being fast and other pools being dense. Ceph clusters are also self-healing, meaning a loss of a drive or even a server can be managed and handled automatically, restoring redundancy to the desired level without user intervention.

This presentation will cover the features of ceph, the basics of how Ceph clusters are architected, and how to configure them for different types of workloads, as well as how the architecture handles scaling and failures.