RAID Redundant Array of Inexpensive Disks
History Single Large Expensive Disk (SLED) Single Large Expensive Disk (SLED) IBM and Berkeley University IBM and Berkeley University RAID levels 0-5 RAID levels 0-5
Why RAID? Redundancy, fault tolerance Redundancy, fault tolerance Increased capacity Increased capacity Increased performance Increased performance
Mirroring The easiest way to get high availability The easiest way to get high availability Half the size Half the size Higher read performance Higher read performance
Striping Higher performance Higher performance Distributed across disks Distributed across disks Work in parallel Work in parallel
Parity Improve the availability Improve the availability Less waste of space Less waste of space
RAID 0 Striping Striping Highest performance Highest performance Lower availability Lower availability
RAID 1 Mirroring Mirroring High availability High availability Low computing power Low computing power
RAID 2, 3, 4 Parity Parity High performance High performance High availability High availability
RAID 5 Parity with distribution Parity with distribution Higher performance than level 1-4 Higher performance than level 1-4 High availability High availability Most used RAID level Most used RAID level
Combinations of different levels Combine two levels and get the advantages from both Combine two levels and get the advantages from both Either hardware driven, software driven or a combination of both Either hardware driven, software driven or a combination of both Examples: 0+1, 1+0, 0+3, 3+0, 0+5, 5+0, 1+5, and 5+1. Examples: 0+1, 1+0, 0+3, 3+0, 0+5, 5+0, 1+5, and 5+1.
Alternatives to RAID Single Large Expensive Disk (SLED) Single Large Expensive Disk (SLED) Redundant Array of Distributed Disks (RADD) Redundant Array of Distributed Disks (RADD) More alternatives: 2D-RADD, C-RAID, RO WB-optimistic and RO WB-pessimistic More alternatives: 2D-RADD, C-RAID, RO WB-optimistic and RO WB-pessimistic
The End First RAID system. Second RAID system.