Shulou(Shulou.com)06/02 Report--

An overview of GFS distributed file system cluster GlusterFS introduction to GlusterFS the open source distributed file system consists of storage servers, clients and NFS/Samba storage gateways without metadata

RDMA: namespace elastic volume management based on the standard protocol GlusterFS term Brick: storage node Volume: volume FUSE: kernel module, client interaction module VFS: virtual port Glusterd: service modular stack architecture modular, stack architecture modular, stack architecture modular

GlusterFS working mode

Application: the client or application accesses the data through the mount point of GlusterFSync. The VFS:linux kernel receives requests and processes FUSE: VFS through VFS API to submit the data to the FUSE kernel file system, fuse

The file system delivers the data to the GlusterFS client GlusterFS Client through the / dev/fuse device file. The data is transferred to the remote GlusterFS Server through the network and written to the server storage device GlusterFS working principle GlusterFS workflow resilient HASH algorithm through the HASH algorithm to get a 32-bit integer divided into N consecutive subspaces, each space corresponds to a Brick elastic HASH algorithm advantages to ensure that the data is evenly distributed in each Brink to solve the metadata server laziness, and then solve the single point of failure and access bottleneck

GlusterFS volumes with four Brick nodes, with an average distribution of range space to the power of 232m

Through the hash algorithm to find the corresponding brick node storage space, to allocate data storage, to call each node data

GlusterFS's volume type distributed stripe volume replication volume distributed stripe replication volume distributed volume without partitioning files by extending file attributes to save hash values support underlying file systems such as ext3, ext4, ZFS, XFS, etc.

Distributed volumes have the following characteristics: files are distributed on different servers. A single point of failure without redundancy to easily and cheaply extend a volume can cause data loss. Create a distributed volume based on lazy data protection to create a distributed volume called dis-volume. The file will be divided into N blocks (N stripe nodes) according to the offset of the gluster volume create dis-volume server1:/dir1 server2:/dir2 stripe volume according to the HASH distributed in server1:/dir1, server2:/dir2 and server3:/dir3. When the polling storage stores large files on each Brick Server node, the performance is particularly outstanding without redundancy, similar to Raid0.

Reading files from multiple server at the same time, efficiency improvement features data are divided into smaller chunks and distributed to different stripes in the block server farm, reducing load and accelerating access speed with smaller files.

No data redundancy

Create a stripe volume create a stripe volume called Stripe-volume, the file will be stored by block polling in Server1:/dir1 and Server2:dir2 Brick, gluster volume create stripe-volume stripe 2 transport tcp server1:/dir1 server2:/dir2 replication volume saves one or more copies of the same file replication mode because the disk utilization is low, the storage space of multiple nodes is inconsistent Then the capacity of the lowest node is taken as the total capacity of the volume according to the bucket effect.

All servers in the characteristic volume keep a complete copy of the volume. The number of copies of the volume can be decided by at least two block servers or more servers when created by the customer.

Have redundancy

Create replication volumes create replication volumes named rep-volume, files will store two replicas at the same time gluster volume create rep-volume replica 2 transport tcp server1:/dir1 server2:/dir2 distributed stripe volumes both distributed volumes and stripe volumes are mainly used for large file access processing

At least 4 servers are required

Create a distributed stripe volume to create a distributed stripe volume called dis-stripe. When configuring a distributed stripe volume, the number of storage servers contained in the Brink in the volume must be a multiple of the stripe number (> = 2) gluster volume create rep-volume stripe 2 transport tcp server1:/dir1 server2:/dir2 server3:/dir1 server4:/dir2 distributed replication volume features both distributed volume and replication volume

For situations where redundancy is required

Create a distributed replication volume create a distributed stripe volume named dis-rep. When configuring a replication volume of a distributed stripe volume, the storage server contained in the Brink in the volume must be a multiple of the stripe number (> = 2 times) gluster volume create rep-volume replica 2 transport tcp server1:/dir1 server2:/dir2 server3:/dir1 server4:/dir2

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.