Diskless shared root cluster

from Wikipedia, the free encyclopedia

A diskless shared root cluster is a system platform for high availability infrastructures.

A Linux cluster file system, for example GFS and OCFS2 , is the basis for connecting a connected storage area network (SAN) to a single system image (SSI) at the file system level.

The architecture of a diskless computer cluster enables the server system and storage array to be separated. Both the operating system and the actual user data (e.g. files, databases or websites) are held centrally and competitively on the attached storage system. A server acting as a cluster node can easily be replaced if necessary.

Although it represents a potential bottleneck, a cluster for simple tasks can also be set up without a SAN by storing the data on an NFS server . However, it is recommended that the NFS server itself is implemented as a 2-node cluster in order to prevent a single point of failure (SPOF).

The infrastructure can be easily scaled thanks to the abstraction between the storage system and the server's computing power. Storage capacity, computing power and network capacity can be adapted to the current requirements independently of one another.

A similar technology can be found in the UNIX area, for example in the TruCluster ( Tru64 UNIX ).

An open source implementation of a diskless shared root cluster is being developed by the Open Sharedroot project.

literature

  • Marc Grimme, Mark Hlawatschek [1] The Diskless Shared-Root Cluster (PDF file; 1.1 MB)
  • Marc Grimme, Mark Hlawatschek, Thomas Merz: [2] Data sharing with a Red Hat GFS storage cluster
  • Kenneth W. Preslan: Red Hat GFS 6.1 - Administrator's Guide

Web links