Catalog

One .GFS summary

1、 file system

①、 File system composition

②、 The role of the file system

③.GFS Professional term

 ④、 GFS Characteristics

Two .GFS working principle

  3、 ... and .GFS Volume type

①. Volume type

---

One .GFS summary

1、 file system

①、 File system composition

•   File system interface （API）
•   A collection of software that manages objects
•   Objects and properties

②、 The role of the file system

•   From a system perspective , File system is to organize and back up the space of file storage device
•   A system that stores and protects and retrieves stored files
•   To be specific , It's responsible for creating files for users 、 Deposit in 、 read out 、 modify 、 Dump files 、 Control the storage of files

③.GFS Professional term

• Brick（ Block storage server ） The server that actually stores user data
• Volume Local file system " Partition "
• FUSE File system in user space （ Category EXT4）,” This is a pseudo file system “, The switching module of the client
• VFS（ Virtual ports ） Kernel virtual file system , The user is submitting a request to VFS then VFS hand FUSH, Give it back GFS client , Finally, the client gives it to the remote storage
• Glusterd（ service ） Is the process that runs the re storage node （ The client is running gluster client）GFS The whole process of use GFS The exchange between is made by Gluster client and glusterd complete

 ④、 GFS Characteristics

• Scalability and high performance ： Extensibility , Expansion nodes , Improve performance through multiple nodes
• High availability ： No single point of failure , There is a backup mechanism , similar Raid The disaster recovery mechanism of
• Global unified namespace ： Centralized management , analogy API The nature of / Concept , The isolation area defined according to his name in the system , It's an independent space ; Unified namespace , Interact with the client , Store the request to the block data server at the back end
• Elastic volume management ： It is convenient for capacity expansion and management and maintenance of back-end storage clusters , More complicated
• Based on standard protocol ： Based on standardized file usage protocol , Give Way CentOS compatible GFS

Two .GFS working principle

• When an external request passes through the mount point ,linux The system kernel passes through VFS Interface , Send a request to FUSE
• FUSE Give the data to the memory /dev/fuse, Then submit to GFS client
• GFS The client processes the data , And through the network protocol （ Such as TCP、IB etc. ）, Transferred to the GFS Server side
• GFS After the server receives the data , adopt VFS Interface , Transfer the data accordingly
   

  3、 ... and .GFS Volume type

• In order to solve the problem of distributed file data index 、 The complexity of positioning , And used HASH Algorithm to assist
• Distributed ( Average distribution ) The benefits of :
• ① When the amount of data is increasing , The amount of data relative to each storage node ( probability ) They are equal.
• ② And if you consider the single point of failure problem , When the data is stored again c Storage nodes , Regarding this GFS There will be a backup mechanism , Default 3 Backup , therefore GFS Its own mechanism will produce redundancy to the data , So as to solve the single point of failure
   

①. Volume type

•   Distributed volumes
•   Strip roll
•   Copy volume
•   Distributed striped volume
•   Distributed replication volumes
•   Strip copy volume
•   Distributed striped data volume

Distributed volumes

File by HASH The algorithm is distributed to all Brick Server On , This kind of roll is GlusterFS The default volume of ; In document units according to HASH The algorithm hashes to different Brick, In fact, it just expands the disk space , If a disk is damaged , Data will also be lost , File level RAID0, No fault tolerance . In this mode , The file is not partitioned , The file is stored directly in some Server Node . Due to the direct use of the local file system for file storage , So access efficiency has not improved , On the contrary, it will be reduced due to network communication .

characteristic

• Files are distributed on different servers , No redundancy
•   Expand the size of the volume more easily and cheaply
•   A single point of failure can cause data loss
•   Rely on the underlying data protection

Strip roll

similar RAID0, Files are divided into data blocks and distributed to multiple servers in a polling manner Brick Server On , File storage is in blocks , Support large file storage , The bigger the file , The more efficient the read is , But there is no redundancy .

characteristic

• The data is divided into smaller pieces and distributed to different stripe areas in the block server cluster
•   Distribution reduces load and smaller files speed up access
•   No data redundancy

Copy volume

Synchronize files to multiple Brick On , Make it have multiple copies of files , It belongs to file level RAID 1, Fault tolerance . Because the data is scattered in multiple Brick in , So the read performance has been greatly improved , But write performance drops . Replication volumes are redundant , Even if one node is damaged , It does not affect the normal use of data . But because you want to save a copy , So disk utilization is low .

characteristic

• All servers in the volume keep a complete copy
•   The number of copies of a volume can be determined when the customer creates
•   At least two block servers or more
•   Redundancy

Distributed striped volume

Brick Server The number is the number of bands （ Block distribution Brick Number ） Multiple , It has the characteristics of distributed roll and strip roll . Mainly used for large file access processing **, Creating a distributed striped volume requires at least 4 Servers .**

Distributed replication volumes

Distributed replication volumes （Distribute Replica volume）：Brick Server The number is the number of mirrors （ Number of data copies ） Multiple , Features of both distributed and replicated volumes , It is mainly used when redundancy is needed .

  Strip copy volume

• Strip copy volume （Stripe Replica volume） similar RAID 1 0, It has the characteristics of striped volume and replicated volume at the same time .
• Distributed striped replication volumes （Distribute Stripe Replicavolume） Composite volume of three basic volumes , Usually used for classes Map Reduce application