Ceph was developed at University of California, Santa Cruz, by Sage Weil in 2003 as a part of his PhD project. The initial implementation provided the Ceph Filesystem (CephFS) in approximately 40,000 lines of C++ code. This was open sourced in 2006 under a Lesser GNU Public License (LGPL) to serve as a reference implementation and research platform. Lawrence Livermore National Laboratory supported Sage's early followup work from 2003 to 2007.

DreamHost, a Los-Angeles-based web hosting and domain registrar company also co-founded by Sage Weil, supported Ceph development from 2007 to 2011. During this period Ceph as we know it took shape: the core components gained stability and reliability, new features were implemented, and the road map for the future was drawn. During this time a number of key developers began contributing, including Yehuda Sadeh-Weinraub, Gregory Farnum, Josh Durgin, Samuel Just, Wido den Hollander, and Loïc Dachary.

In 2012 Sage Weil founded Inktank to enable the widespread adoption of Ceph. Their expertise, processes, tools, and support enabled enterprise-subscription customers to effectively implement and manage Ceph storage systems. In 2014 Red Hat, Inc.,the world's leading provider of open source solutions, agreed to acquire Inktank.

The term Ceph is a common nickname given to pet octopuses; Ceph and is an abbreviation of cephalopod, marine animals belonging to the Cephalopoda class of molluscs. Ceph's mascot is an octopus,referencing the highly parallel behavior of an octopus and was chosen to connect the file system with UCSC's mascot, a banana slug named Sammy. Banana slugs are gastropods,which are also a class of molluscs. As Ceph is not an acronym, it should not be uppercased as CEPH.

Each release of Ceph has a numeric version. Major releases also receive cephalopod code-names in alphabetical order. Through the Luminous release the Ceph community tagged a new major version about twice a year, alternating between Long Term Support (LTS) and stable releases. The latest two LTS releases were officially supported, but only the single latest stable release.

The release numbering scheme has changed since the first edition of Learning Ceph was published. Earlier major releases were tagged initially with a version number (0.87) and were followed by multiple point releases (0.87.1, 0.87.2, ...). Releases beginning with Infernalis however are numbered as shown:

The major release number matches the letter of the alphabet of its code name (for example I is the ninth letter of the English alphabet, so 9.2.1 was named Infernalis). As we write, there have been four releases following this numbering convention: Infernalis, Jewel, Kraken, and Luminous.

The early versions of each major release have a type of 0 in the second field, which indicates active pre-release development status for early testers and the brave of heart. Later release candidates have a type of 1 and are targeted at test clusters and brave users. A type of 2 represents a general-availability, production-ready release. Point releases mostly contain security and bug fixes, but sometimes offer functionality improvements as well.

The Jewel LTS release brought a number of significant changes:

• Unified queue of client I/O, recovery, scrubs, and snapshot trimming
• Daemons now run as the ceph user, which must be addressed when upgrading
• Cache tier improvements
• SHEC erasure coding is no longer experimental
• The SWIFT API now supports object expiration
• RBD improvements (now supports suffixes)
• rbd du shows actual and provisioned usage quickly via object-map and fast-diff features
• New rbd status command
• deep-flatten now handles snapshots
• CephFS snapshots can now be renamed
• And CephFS is considered stable!
• Scrubbing improvements
• TCMalloc improvements
• Multisite functionality in RGW significantly improved
• OpenStack Keystone v3 support
• Swift per-tenant namespace
• Async RBD mirroring
• A new look for ceph status

As we write, the major Luminous LTS release has just reached general availability. Early experiences are positive and it is the best choice for new deployments. Much-anticipated features in Luminous include:

• The BlueStore back end is supported
• In-line compression and read checksums
• Erasure coding for RBD volumes
• Better tools for uniform OSD utilization
• Improved tools for the OSD lifecycle
• Enhanced CLI
• Multiple active CephFS MDS servers are supported