RAID - jude-lindale/Wiki GitHub Wiki

Why we need RAID?

• hard disks are slow/ slow access time
• prone to fail

What is RAID?

• redundant array of independent disk
• many levels of redundancy

Types

• RAID 0:
  • files are divided into two halves and each half is stored on different disks from the other
  • increase performance
  • time is reduced
  • two or more disks
  • no redundancy
  • no error checking
  • increased read and write speed
• RAID 1:
  • duplicates data over two or more disks/mirroring
  • if data is lost there is always a backup
  • no error checking
• RAID 2:
  • has stripping
  • checking and correcting / parity
  • minimum of 3 disks at the bit level
  • all disks are spinning in unison
  • redundancy by using ECC bits on different disks
  • lots of disks required for ECC
  • do not use
  • uses Hamming code to detect and correct errors
• RAID 3:
  • still not fast and not much better than RIAD 2
  • only can correct data using XOR (communicative relation)
  • do not use
  • all disks are spinning in unison
  • A XOR B = C
  • B XOR C = A
  • C XOR C = B
  • works with any number of inputs
  • 01101000 01101001 (plain)
  • 01010010 01000101 (key)
  • 00111010 00101100 (result/ parity disk)
  • 01101000 01101001 + 01010010 01000101 = 00111010 00101100
  • 01010010 01000101 + 00111010 00101100 = 01101000 01101001
• RIAD 4:
  • minimum of 3 disks
  • faster than RAID 3
  • does not need to spin in unison
  • is not used because each time a disk is changed the parity disk will change
  • parity more prone to failure
  • do not use
• RAID 5:
  • parities are spread over multiple disks
  • if more than one disk fails then data cannot be recovered
  • most common RAID configuration
• RAID 6:
  • same as RAID 5 but there are two copies of parities over multiple disks
  • not very common
  • if more than two disks fail then they can be recovered