Managing software RAID in Linux system

Linux systems can efficiently handle software RAID systems. Before going in how Linux handles RAID system lets see what is RAID and what different levels of RAID are available.

What is RAID

RAID stands for Redundant Array of Inexpensive Disks. In a RAID system data is stored in multiple disk drives considering it as single logical drive. The data can be stored in several ways depending upon level of redundancy (referred as RAID level) and performance.

Levels of RAID

RAID 0 (Stripping)

RAID level 0 uses stripping without mirroring or parity. Data is written on Disk in sequence, this results in excellent performance of System. But there is no redundancy i.e. in case of one disk fails complete system will fail. Hundred percent of Data storage can be used, if disks of similar size are used.  This RAID system should be used where data is not so critical or is regularly backed up since RAID 0 system has no fault tolerance.

Stripping of data in RAID 0 system means data is broken into chunks or pieces or strips of size, predefined by user, and in parallel written on different disk sets. This RAID system is faster because data is written simultaneously on all disks.

Features

  • Stripping of Data is used in RAID 0 system
  • Fast read and write performance
  • Capacity of RAID system = storage capacity of Disk with lowest storage capacity*number of disks
  • Minimum 2 drives are required for RAID 0 system

Disadvantage

  • No redundancy i.e. in case of any disk failure complete system fails

RAID 1 (Mirroring)

RAID 1 or RAID Level 1 employs mirroring technology. This means redundancy of storage is there but this reduces usable storage space by half. This RAID level should be used where fault tolerance is more important than performance. In case of one of the disk fails other working disk is used by system as independent disk. Once faulty drive is replaced it is rebuilt as and RAID system is again functional.

Mirroring of data in RAID 1 systems means identical data is written on all RAID member disks at same. This ensure you have complete backup at all the time. Data write is slow as data has is written across all disks simultaneously but read is always faster as all disks are accessed in parallel.

Features

  • Mirroring of Data is used in RAID 1 system.
  • Faster read as disk are accessed in parallel for data but slower write performance as data is to be written simultaneously across all member disks.
  • Capacity of RAID system = storage capacity of Disk with lowest storage capacity.
  • Minimum 2 drives are required for RAID 1 system. Using more than two drives given no additional advantage.
  • High redundancy as back up is always available.

Disadvantage

  • Slow write speed.

RAID 5 (stripping with parity of data)

In RAID level 5 data stripping is done alone with a parity.

What is parity? RAID level 5 uses parity data for recovery as well as provide data for read request in case one of the disk fails. This parity is checksum of all data written.

The parity data is stripped across disks. In case of any of the disk fails. This parity data can be used to reconstruct data in case one of the disk fails. Also satisfy all I/O request to RAID system while data is being rebuilt.

Features

  • Stripping of data with parity is used in RAID 5 system.
  • Faster read as disk are accessed in parallel for data but slower write performance parity data calculation overhead for written data.
  • Capacity of RAID system = (storage capacity of Disk with lowest storage capacity)*(Number of drives – 1).
  • Minimum 3 drives are required for RAID 5 system. Using more than 3 drives improves efficiency of RAID 5 system.
  • High redundancy.

Disadvantage

  • Slow write speed due calculation of parity data during write operation.

RAID 10  (stripped mirror)

RAID level 10 uses both stripping as well as mirroring to store data. The system is very efficient when performing data read/write operations. In case of RAID 10 there is cluster of 2 disks and such cluster sets combine to form an array of disks. In RAID 10 data is first stripped array of disk and this stripped data is then mirrored on each set of cluster. This RAID system should be used where data is mission critical. Also this RAID level is not cost effective as half of space is lost to mirroring.

Stripping of data in RAID 0 system means data is broken into chunks or pieces or strips of size, predefined by user, and in parallel written on different disk sets. This RAID system is faster because data is written simultaneously on all disks.

Features

  • Stripped mirror of Data is used in RAID 10 system
  • Fast read and write performance
  • Capacity of RAID system = (storage capacity of Disk with lowest storage capacity*number of disks)/2
  • Minimum 4 drives are required for RAID 0 system.
  • Excellent redundancy which improves with increase in number of disks. Important when increasing number of disk remember every increase will require a set of two disks.

Disadvantage

  • High overhead cost as requires higher number of disks.

For remaining article wait for next update……

Next coming soon