GlusterFS Gluster
glusterfs scale-out network-attached storage file system. has found applications including cloud computing, streaming media services, , content delivery networks. glusterfs developed gluster, inc. , red hat, inc., result of red hat acquiring gluster in 2011.
in june 2012, red hat storage server announced commercially supported integration of glusterfs red hat enterprise linux. red hat bought inktank storage in april 2014, company behind ceph distributed file system, , re-branded glusterfs-based red hat storage server red hat gluster storage .
design
glusterfs aggregates various storage servers on ethernet or infiniband rdma interconnect 1 large parallel network file system. free software, parts licensed under gnu general public license (gpl) v3 while others dual licensed under either gpl v2 or lesser general public license (lgpl) v3. glusterfs based on stackable user space design.
glusterfs has client , server component. servers typically deployed storage bricks, each server running glusterfsd daemon export local file system volume. glusterfs client process, connects servers custom protocol on tcp/ip, infiniband or sockets direct protocol, creates composite virtual volumes multiple remote servers using stackable translators. default, files stored whole, striping of files across multiple remote volumes supported. final volume may mounted client host using own native protocol via fuse mechanism, using nfs v3 protocol using built-in server translator, or accessed via gfapi client library. native-protocol mounts may re-exported e.g. via kernel nfsv4 server, samba, or object-based openstack storage (swift) protocol using ufo (unified file , object) translator.
most of functionality of glusterfs implemented translators, including file-based mirroring , replication, file-based striping, file-based load balancing, volume failover, scheduling , disk caching, storage quotas, , volume snapshots user serviceability (since glusterfs version 3.6).
the glusterfs server intentionally kept simple: exports existing directory as-is, leaving client-side translators structure store. clients stateless, not communicate each other, , expected have translator configurations consistent each other. glusterfs relies on elastic hashing algorithm, rather using either centralized or distributed metadata model. version 3.1 , later of glusterfs, volumes can added, deleted, or migrated dynamically, helping avoid configuration coherency problems, , allowing glusterfs scale several petabytes on commodity hardware avoiding bottlenecks affect more tightly coupled distributed file systems.
glusterfs has been used foundation academic research , survey article.
red hat markets software 3 markets: on-premises , public cloud , private cloud .
Comments
Post a Comment