Gluster Inc. (formerly known as Z RESEARCH^[1][2][3]) was asoftware company that provided anopen source platform forscale-out public and privatecloud storage. The company was privately funded and headquartered inSunnyvale, California, with an engineering center inBangalore, India. Gluster was funded by Nexus Venture Partners andIndex Ventures. Gluster was acquired byRed Hat on October 7, 2011.^[4]

Gluster, Inc.

Company type	Privately funded
Industry	Software,computer storage
Founded	2005
Founder	Anand Avati Anand Babu Periasamy
Headquarters	Sunnyvale, California andBangalore,India
Number of locations	2
Key people	Anand Babu (AB) Periasamy (CTO) and Hitesh Chellani (CEO)
Products	Cloud storage
Number of employees	60
Website	www.gluster.com

The nameGluster comes from the combination of the termsGNU andcluster.^[2] Despite the similarity in names, Gluster is not related to theLustre file system and does not incorporate any Lustre code.Gluster based its product onGlusterFS, an open-source software-basednetwork-attachedfilesystem that deploys on commodity hardware.^[5] The initial version of GlusterFS was written by Anand Babu Periasamy, Gluster's founder and CTO.^[6]In May 2010 Ben Golub became the president and chief executive officer.^[7][8]

Red Hat became the primary author and maintainer of the GlusterFSopen-source project after acquiring the Gluster company in October 2011.^[4]The product was first marketed as Red Hat Storage Server, but in early 2015 renamed to be Red Hat Gluster Storage since Red Hat has also acquired theCephfile system technology.^[9]

Red Hat Gluster Storage is in the retirement phase of its lifecycle with a end of support life date of December 31, 2024.^[10]

The GlusterFS architecture aggregates compute, storage, and I/O resources into a global namespace. Each server plus attached commodity storage (configured asdirect-attached storage,JBOD, or using astorage area network) is considered to be a node. Capacity is scaled by adding additional nodes or adding additional storage to each node. Performance is increased by deploying storage among more nodes. High availability is achieved by replicating data n-way between nodes.

For public cloud deployments, GlusterFS offers anAmazon Web Services (AWS)Amazon Machine Image (AMI), which is deployed on Elastic Compute Cloud (EC2) instances rather than physical servers and the underlying storage is Amazon'sElastic Block Storage (EBS).^[11] In this environment, capacity is scaled by deploying more EBS storage units, performance is scaled by deploying more EC2 instances, and availability is scaled by n-way replication between AWS availability zones.

A typical on-premises, or private cloud deployment will consist of GlusterFS installed as a virtual appliance on top of multiple commodity servers runninghypervisors such asKVM,Xen, orVMware; or on bare metal.^[12]

GlusterFS is ascale-outnetwork-attached storagefile system. It has found applications includingcloud computing, streaming media services, and content delivery networks. GlusterFS was developed originally by Gluster, Inc. and then byRed Hat, Inc., as a result of Red Hat acquiring Gluster in 2011.^[15]

In June 2012,Red Hat Storage Server was announced as a commercially supported integration of GlusterFS withRed Hat Enterprise Linux.^[16] Red Hat boughtInktank Storage in April 2014, which is the company behind theCeph distributed file system, and re-branded GlusterFS-based Red Hat Storage Server to "Red Hat Gluster Storage".^[17]

GlusterFS aggregates various storage servers overEthernet orInfinibandRDMA interconnect into one large parallel network file system. It is free software, with some parts licensed under the GNUGeneral Public License (GPL) v3 while others are dual licensed under either GPL v2 or theLesser General Public License (LGPL) v3. GlusterFS is based on a stackable user space design.

GlusterFS has a client and server component. Servers are typically deployed asstorage bricks, with each server running aglusterfsd daemon to export a local file system as avolume. Theglusterfs client process, which connects to servers with a custom protocol overTCP/IP, InfiniBand orSockets Direct Protocol, creates composite virtual volumes from multiple remote servers using stackabletranslators. By default, files are stored whole, butstriping of files across multiple remote volumes is also possible. The client maymount the composite volume using a GlusterFS native protocol via theFUSE mechanism or usingNFSv3 protocol using a built-in server translator, or access the volume via thegfapi client library. The client may re-export a native-protocol mount, for example via the kernelNFSv4 server,SAMBA, or the object-basedOpenStack Storage (Swift) protocol using the "UFO" (Unified File and Object) translator.

Most of the functionality of GlusterFS is implemented as translators, including file-basedmirroring andreplication, file-basedstriping, file-basedload balancing, volumefailover,scheduling anddisk caching,storage quotas, and volumesnapshots with user serviceability (since GlusterFS version 3.6).

The GlusterFS server is intentionally kept simple: it exports an existingdirectory as-is, leaving it up to client-side translators to structure the store. The clients themselves are stateless, do not communicate with each other, and are expected to have translator configurations consistent with each other. GlusterFS relies on an elastichashing algorithm, rather than using either a centralized or distributed metadata model. The user can add, delete, or migrate volumes dynamically, which helps to avoid configurationcoherency problems. This allows GlusterFS to scale up to severalpetabytes oncommodity hardware by avoiding bottlenecks that normally affect more tightly coupled distributed file systems.

GlusterFS provides data reliability and availability through various kinds of replication: replicated volumes andgeo-replication.^[18] Replicated volumes ensure that there exists at least one copy of each file across the bricks, so if one fails, data is still stored and accessible. Geo-replication provides a leader-follower model of replication, where volumes are copied across geographically distinct locations. This happens asynchronously and is useful for availability in case of a whole data center failure.

GlusterFS has been used as the foundation for academic research^[19][20]and a survey article.^[21]

Red Hat markets the software for three markets: "on-premises",public cloud and "private cloud".^[22]