Ceph Glossary

Application

More properly called a client, an application is any program external to Ceph that uses a Ceph Cluster to store and replicate data.

BlueStore 

OSD BlueStore is a storage back end used by OSD daemons, and was designed specifically for use with Ceph. BlueStore was introduced in the Ceph Kraken release. The Luminous release of Ceph promoted BlueStore to the default OSD back end, supplanting FileStore. As of the Reef release, FileStore is no longer available as a storage back end.

BlueStore stores objects directly on raw block devices or partitions, and does not interact with mounted file systems. BlueStore uses RocksDB’s key/value database to map object names to block locations on disk.

Bucket

In the context of RGW, a bucket is a group of objects. In a filesystem-based analogy in which objects are the counterpart of files, buckets are the counterpart of directories. Multisite sync policies can be set on buckets, to provide fine-grained control of data movement from one zone to another zone.

The concept of the bucket has been taken from AWS S3. See also the AWS S3 page on creating buckets and the AWS S3 ‘Buckets Overview’ page.

OpenStack Swift uses the term “containers” for what RGW and AWS call “buckets”. See the OpenStack Storage API overview page.

Ceph

Ceph is a distributed network storage and file system with distributed metadata management and POSIX semantics.

ceph-ansible 

A GitHub repository, supported from the Jewel release to the Quincy release, that facilitates the installation of a Ceph cluster.

Ceph Block Device

Also called “RADOS Block Device” and RBD. A software instrument that orchestrates the storage of block-based data in Ceph. Ceph Block Device splits block-based application data into “chunks”. RADOS stores these chunks as objects. Ceph Block Device orchestrates the storage of those objects across the storage cluster.

Ceph Block Storage

One of the three kinds of storage supported by Ceph (the other two are object storage and file storage). Ceph Block Storage is the block storage “product”, which refers to block-storage related services and capabilities when used in conjunction with the collection of (1) librbd (a python module that provides file-like access to RBD images), (2) a hypervisor such as QEMU or Xen, and (3) a hypervisor abstraction layer such as libvirt.

Ceph Client

Any of the Ceph components that can access a Ceph Storage Cluster. This includes the Ceph Object Gateway, the Ceph Block Device, the Ceph File System, and their corresponding libraries. It also includes kernel modules, and FUSEs (Filesystems in USERspace).

Ceph Client Libraries

The collection of libraries that can be used to interact with components of the Ceph Cluster.

Ceph Cluster Map

See Cluster Map

Ceph Dashboard

The Ceph Dashboard is a built-in web-based Ceph management and monitoring application through which you can inspect and administer various resources within the cluster. It is implemented as a Ceph Manager Daemon module.

Ceph File System

See CephFS

CephFS 

The Ceph File System, or CephFS, is a POSIX-compliant file system built on top of Ceph’s distributed object store, RADOS. See CephFS Architecture for more details.

ceph-fuse 

ceph-fuse is a FUSE (”Filesystem in USErspace”) client for CephFS. ceph-fuse mounts a Ceph FS ata specified mount point.

Ceph Interim Release

See Releases.

Ceph Kernel Modules

The collection of kernel modules that can be used to interact with the Ceph Cluster (for example: ceph.ko, rbd.ko).

Ceph Manager 

The Ceph manager daemon (ceph-mgr) is a daemon that runs alongside monitor daemons to provide monitoring and interfacing to external monitoring and management systems. Since the Luminous release (12.x), no Ceph cluster functions properly unless it contains a running ceph-mgr daemon.

Ceph Manager Dashboard

See Ceph Dashboard.

Ceph Metadata Server

See MDS.

Ceph Monitor

A daemon that maintains a map of the state of the cluster. This “cluster state” includes the monitor map, the manager map, the OSD map, and the CRUSH map. A Ceph cluster must contain a minimum of three running monitors in order to be both redundant and highly-available. Ceph monitors and the nodes on which they run are often referred to as “mon”s. See Monitor Config Reference.

Ceph Node

A Ceph node is a unit of the Ceph Cluster that communicates with other nodes in the Ceph Cluster in order to replicate and redistribute data. All of the nodes together are called the Ceph Storage Cluster. Ceph nodes include OSDs, Ceph Monitors, Ceph Managers, and MDSes. The term “node” is usually equivalent to “host” in the Ceph documentation. If you have a running Ceph Cluster, you can list all of the nodes in it by running the command ceph node ls all.

Ceph Object Gateway 

An object storage interface built on top of librados. Ceph Object Gateway provides a RESTful gateway between applications and Ceph storage clusters.

Ceph Object Storage

See Ceph Object Store.

Ceph Object Store

A Ceph Object Store consists of a Ceph Storage Cluster and a Ceph Object Gateway (RGW).

Ceph OSD 

Ceph Object Storage Daemon. The Ceph OSD software, which interacts with logical disks (OSD). Around 2013, there was an attempt by “research and industry” (Sage’s own words) to insist on using the term “OSD” to mean only “Object Storage Device”, but the Ceph community has always persisted in using the term to mean “Object Storage Daemon” and no less an authority than Sage Weil himself confirms in November of 2022 that “Daemon is more accurate for how Ceph is built” (private correspondence between Zac Dover and Sage Weil, 07 Nov 2022).

Ceph OSD Daemon

See Ceph OSD.

Ceph OSD Daemons

See Ceph OSD.

Ceph Platform

All Ceph software, which includes any piece of code hosted at https://github.com/ceph.

Ceph Point Release

See Releases.

Ceph Project

The aggregate term for the people, software, mission and infrastructure of Ceph.

Ceph Release

See Releases.

Ceph Release Candidate

See Releases.

Ceph Stable Release

See Releases.

Ceph Stack

A collection of two or more components of Ceph.

Ceph Storage Cluster 

The collection of Ceph Monitors, Ceph Managers, Ceph Metadata Servers, and OSDs that work together to store and replicate data for use by applications, Ceph Users, and Ceph Clients. Ceph Storage Clusters receive data from Ceph Clients.

CephX

The Ceph authentication protocol. CephX authenticates users and daemons. CephX operates like Kerberos, but it has no single point of failure. See the High-availability Authentication section of the Architecture document and the CephX Configuration Reference.

Client

A client is any program external to Ceph that uses a Ceph Cluster to store and replicate data.

Cloud Platforms

Cloud Stacks

Third party cloud provisioning platforms such as OpenStack, CloudStack, OpenNebula, and Proxmox VE.

Cluster Map

The set of maps consisting of the monitor map, OSD map, PG map, MDS map, and CRUSH map, which together report the state of the Ceph cluster. See the “Cluster Map” section of the Architecture document for details.

Crimson

A next-generation OSD architecture whose core aim is the reduction of latency costs incurred due to cross-core communications. A re-design of the OSD that reduces lock contention by reducing communication between shards in the data path. Crimson improves upon the performance of classic Ceph OSDs by eliminating reliance on thread pools. See Crimson: Next-generation Ceph OSD for Multi-core Scalability. See the Crimson developer documentation.

CRUSH

Controlled Replication Under Scalable Hashing. The algorithm that Ceph uses to compute object storage locations. See CRUSH: Controlled, Scalable, Decentralized Placement of Replicated Data.

CRUSH rule

The CRUSH data placement rule that applies to a particular pool or pools.

DAS

Direct-Attached Storage. Storage that is attached directly to the computer accessing it, without passing through a network. Contrast with NAS and SAN.

Dashboard 

A built-in web-based Ceph management and monitoring application to administer various aspects and objects of the cluster. The dashboard is implemented as a Ceph Manager module. See Ceph Dashboard for more details.

Dashboard Module

Another name for Dashboard.

Dashboard Plugin

The dashboard plugin was a Mimic-era web application that visualized information and statistics about the Ceph cluster using a web server hosted by the Ceph Manager.

See the Mimic-era Dashboard Plugin documentation.

DC

Data Center.

Flapping OSD

An OSD that is repeatedly marked up and then down in rapid succession. See Flapping OSDs.

FQDN

Fully Qualified Domain Name. A domain name that is applied to a node in a network and that specifies the node’s exact location in the tree hierarchy of the DNS.

In the context of Ceph cluster administration, FQDNs are often applied to hosts. In this documentation, the term “FQDN” is used mostly to distinguish between FQDNs and relatively simpler hostnames, which do not specify the exact location of the host in the tree hierarchy of the DNS but merely name the host.

Host

Any single machine or server in a Ceph Cluster. See Ceph Node.

Hybrid OSD

Refers to an OSD that has both HDD and SSD drives.

librados

An API that can be used to create a custom interface to a Ceph storage cluster. librados makes it possible to interact with Ceph Monitors and with OSDs. See Introduction to librados. See librados (Python).

LVM tags

Logical Volume Manager tags. Extensible metadata for LVM volumes and groups. They are used to store Ceph-specific information about devices and its relationship with OSDs.

MDS

The Ceph MetaData Server daemon. Also referred to as “ceph-mds”. The Ceph metadata server daemon must be running in any Ceph cluster that runs the CephFS file system. The MDS stores all filesystem metadata. Clients work together with either a single MDS or a group of MDSes to maintain a distributed metadata cache that is required by CephFS.

See Deploying Metadata Servers.

See the ceph-mds man page.

MGR

The Ceph manager software, which collects all the state from the whole cluster in one place.

MON 

The Ceph monitor software.

Monitor Store

The persistent storage that is used by the Monitor. This includes the Monitor’s RocksDB and all related files in /var/lib/ceph.

Node

See Ceph Node.

Object Storage

Object storage is one of three kinds of storage relevant to Ceph. The other two kinds of storage relevant to Ceph are file storage and block storage. Object storage is the category of storage most fundamental to Ceph.

Object Storage Device

See OSD.

OMAP

“object map”. A key-value store (a database) that is used to reduce the time it takes to read data from and to write to the Ceph cluster. RGW bucket indexes are stored as OMAPs. Erasure-coded pools cannot store RADOS OMAP data structures.

Run the command ceph osd df to see your OMAPs.

See Eleanor Cawthon’s 2012 paper A Distributed Key-Value Store using Ceph (17 pages).

OpenStack Swift

In the context of Ceph, OpenStack Swift is one of the two APIs supported by the Ceph Object Store. The other API supported by the Ceph Object Store is S3.

See the OpenStack Storage API overview page.

OSD

Probably Ceph OSD, but not necessarily. Sometimes (especially in older correspondence, and especially in documentation that is not written specifically for Ceph), “OSD” means “Object Storage Device”, which refers to a physical or logical storage unit (for example: LUN). The Ceph community has always used the term “OSD” to refer to Ceph OSD Daemon despite an industry push in the mid-2010s to insist that “OSD” should refer to “Object Storage Device”, so it is important to know which meaning is intended.

OSD, flapping

See Flapping OSD.

OSD FSID

The OSD fsid is a unique identifier that is used to identify an OSD. It is found in the OSD path in a file called osd_fsid. The term FSID is used interchangeably with UUID.

OSD ID

The OSD id an integer unique to each OSD (each OSD has a unique OSD ID). Each OSD id is generated by the monitors during the creation of its associated OSD.

OSD UUID

The OSD UUID is the unique identifier of an OSD. This term is used interchangeably with FSID.

Period

In the context of RGW, a period is the configuration state of the Realm. The period stores the configuration state of a multi-site configuration. When the period is updated, the “epoch” is said thereby to have been changed.

Placement Groups (PGs)

Placement groups (PGs) are subsets of each logical Ceph pool. Placement groups perform the function of placing objects (as a group) into OSDs. Ceph manages data internally at placement-group granularity: this scales better than would managing individual (and therefore more numerous) RADOS objects. A cluster that has a larger number of placement groups (for example, 100 per OSD) is better balanced than an otherwise identical cluster with a smaller number of placement groups.

Ceph’s internal RADOS objects are each mapped to a specific placement group, and each placement group belongs to exactly one Ceph pool.

PLP

Power Loss Protection. A technology that protects the data of solid-state drives by using capacitors to extend the amount of time available for transferring data from the DRAM cache to the SSD’s permanent memory. Consumer-grade SSDs are rarely equipped with PLP.

Pool 

A pool is a logical partition used to store objects.

Pools

See pool.

Primary Affinity 

The characteristic of an OSD that governs the likelihood that a given OSD will be selected as the primary OSD (or “lead OSD”) in an acting set. Primary affinity was introduced in Firefly (v. 0.80). See Primary Affinity.

Prometheus 

An open-source monitoring and alerting toolkit. Ceph offers a “Prometheus module”, which provides a Prometheus exporter that passes performance counters from a collection point in ceph-mgr to Prometheus.

Quorum

Quorum is the state that exists when a majority of the Monitors in the cluster are up. A minimum of three Monitors must exist in the cluster in order for Quorum to be possible.

RADOS

Reliable Autonomic Distributed Object Store. RADOS is the object store that provides a scalable service for variably-sized objects. The RADOS object store is the core component of a Ceph cluster. This blog post from 2009 provides a beginner’s introduction to RADOS. Readers interested in a deeper understanding of RADOS are directed to RADOS: A Scalable, Reliable Storage Service for Petabyte-scale Storage Clusters.

RADOS Cluster

A proper subset of the Ceph Cluster consisting of OSDs, Ceph Monitors, and Ceph Managers.

RADOS Gateway

See RGW.

RBD

RADOS Block Device. See Ceph Block Device.

Realm 

In the context of RADOS Gateway (RGW), a realm is a globally unique namespace that consists of one or more zonegroups.

Releases

Ceph Interim Release: A version of Ceph that has not yet been put through quality assurance testing. May contain new features.
Ceph Point Release: Any ad hoc release that includes only bug fixes and security fixes.
Ceph Release: Any distinct numbered version of Ceph.
Ceph Release Candidate: A major version of Ceph that has undergone initial quality assurance testing and is ready for beta testers.
Ceph Stable Release: A major version of Ceph where all features from the preceding interim releases have been put through quality assurance testing successfully.

Reliable Autonomic Distributed Object Store

The core set of storage software which stores the user’s data (MON+OSD). See also RADOS.

RGW 

RADOS Gateway.

Also called “Ceph Object Gateway”. The component of Ceph that provides a gateway to both the Amazon S3 RESTful API and the OpenStack Swift API.

S3

In the context of Ceph, S3 is one of the two APIs supported by the Ceph Object Store. The other API supported by the Ceph Object Store is OpenStack Swift.

See the Amazon S3 overview page.

scrubs

The processes by which Ceph ensures data integrity. During the process of scrubbing, Ceph generates a catalog of all objects in a placement group, then ensures that none of the objects are missing or mismatched by comparing each primary object against its replicas, which are stored across other OSDs. Any PG is determined to have a copy of an object that is different than the other copies or is missing entirely is marked “inconsistent” (that is, the PG is marked “inconsistent”).

There are two kinds of scrubbing: light scrubbing and deep scrubbing (also called “normal scrubbing” and “deep scrubbing”, respectively). Light scrubbing is performed daily and does nothing more than confirm that a given object exists and that its metadata is correct. Deep scrubbing is performed weekly and reads the data and uses checksums to ensure data integrity.

See Scrubbing in the RADOS OSD Configuration Reference Guide and page 141 of Mastering Ceph, second edition (Fisk, Nick. 2019).

secrets

Secrets are credentials used to perform digital authentication whenever privileged users must access systems that require authentication. Secrets can be passwords, API keys, tokens, SSH keys, private certificates, or encryption keys.

SDS

Software-defined Storage.

systemd oneshot

A systemd type where a command is defined in ExecStart which will exit upon completion (it is not intended to daemonize)

Swift

See OpenStack Swift.

Teuthology

The collection of software that performs scripted tests on Ceph.

User

An individual or a system actor (for example, an application) that uses Ceph clients to interact with the Ceph Storage Cluster. See User and User Management.

Zone

In the context of RGW, a zone is a logical group that consists of one or more RGW instances. A zone’s configuration state is stored in the period. See Zones.