GlusterFS

from Wikipedia, the free encyclopedia

GlusterFS is a distributed file system that presents storage elements from multiple servers as a unified file system. The various servers, including cluster -nodes ( English node , node ') called, form a client-server architecture via TCP / IP . As a special feature, NAS systems can be integrated directly into the cluster via Infiniband , and a redundant connection of storage media via TCP / IP, Infiniband Verbs or InfiniBand SDP (Socket Direct Protocol) is also possible. The data on all cluster nodes can be read and written at the same time, with any changes to files on all servers being implemented immediately. The file system is a FUSE integrated kernel module and is used by POSIX support -enabled operating systems, including Linux , FreeBSD , OpenSolaris and Mac OS X . No kernel module is required to start a GlusterFS server. A server can be both a client and a server at the same time. A client for Windows systems is being planned, but the developers will not implement it until the WinFUSE project is stable.

The development of GlusterFS began in mid-2005 by the GlusterOS development team from Z Research Inc., a first release of the file system was published in July 2006. GlusterFS is licensed under the GPL in version 3. The developers offer paid support. In late 2011, Red Hat bought GlusterFS for $ 136 million.

Functionality

GlusterFS has a modular design and supports several operating modes:

Standalone storage
a single server that provides the file system over the network (similar to NFS )
Distributed storage
several servers that store the data distributed among each other and make it available to the clients
Replicated storage
several servers that store the data in a mirrored manner and make them available to the clients
Distributed replicated storage
several servers that distribute and mirror the data among each other
Striped storage
multiple servers where the data is stripped in order to provide higher performance and disk IO bandwidth
Cloud / HPC storage
see Distributed Replicated Storage
NFS-like standalone storage server-2
Similar to standalone storage, it provides more than just a file system
Aggregating Three Storage Servers with Unify
three servers that provide a uniform file system using Unify, without redundancy

Modular construction

GlusterFS uses a modular structure so that all functions described can be mapped with just a few components. The components are data partitions (volumes), transport groups and translators. The translators offer the possibility of adding functions to the file system, in particular whether it should communicate via Infiniband, TCP or a mixture of both, which physical partitions or directories should be integrated and which RAID levels should be used. To illustrate the modular concept, there is a ROT13 translator for "encrypting" the file system.

Possible applications

With GlusterFS a kind of network RAID can be created from which several computers can access a common file system at the same time. There are no limitations here, such as being able to use a maximum of two servers. GlusterFS is fault-tolerant , since with GlusterFS user data , metadata and namespace can be stored in a distributed manner. Each additional GlusterFS server increases the maximum data throughput of the file system, so that I / O bandwidths of a few gigabytes per second can be achieved here.

For processors (CPUs) that applies Moore's Law , but which does not apply in storage media and storage solutions, although there is also a need for larger and faster memory. Often it is not the CPU performance of a server that is the bottleneck , but more and more often the system's data storage device which is too slow. GlusterFS provides a remedy here with the ability to scale as required .

See also

Web links

Individual evidence

  1. ↑ " Red Hat is buying storage specialist Gluster ", linux-magazin, October 4th, 2011