“Ceph is a unified, distributed storage system designed for excellent performance, reliability and scalability.”
Our journey with Ceph began in 2013 when we needed image and volume storage for our openstack based private cloud. Using the “ceph block storage”, it’s possible to utilize lots of different disks spread over different hosts without any single point of failure.
Recently, it’s becoming more popular to use that for persistent volumes in containerized environments, specifically kubernetes. Using rook, deploying a Ceph cluster and provisioning volumes for statefulsets is only a few steps away.
On the other hand Ceph is an object storage system at heart, and it’s only logical to use it as one. With S3 or Swift compatible API, it’s widely deployed as a general purpose object storage for media files, backups, application packages or big data systems.
It’s even possible to use the same cluster for volume, object and file system needs,… which we did. Though in practice, it’s pretty rare to use the same physical cluster for different requirements. Virtual machines generally benefit from high levels of iops capabilities. An archival object storage will prioritize capacity efficiency. Nevertheless the same infrastructure for different types of storage means extreme flexibility and fewer technologies to support.