RAID.htm

Redundant Array of Independent Disks

Key Terms

• Parity: Parity is a method used in some RAID levels (like RAID 5 and 6) to provide fault tolerance. It's a calculated value that can be used to reconstruct data if a drive fails.
• Stripe Size: Stripe size refers to the amount of data written to each disk in a RAID array before moving to the next disk.

RAID is a storage technology that combines multiple disk drive components into a logical unit for data redundancy and performance improvement.

RAID works by distributing data across multiple drives in one of several ways, depending on the RAID level. The most common methods are:

1. Striping: Splitting data across multiple drives
2. Mirroring: Duplicating data on separate drives
3. Parity: Storing calculated recovery information

Key RAID levels:

• RAID 0 (Striping):
  • Data is split across drives
  • Improves performance, no redundancy
  • Example: A 100MB file on 2 drives might use 50MB on each
• RAID 1 (Mirroring):
  • Data is duplicated on separate drives
  • Provides redundancy, slightly improves read performance
  • Example: A 100MB file is written identically to 2 drives
• RAID 5 (Striping with distributed parity):
  • Data and parity are distributed across all drives
  • Good balance of performance and redundancy
  • Example: In a 4-drive array, 3 parts data + 1 part parity per stripe, rotating parity location
• RAID 6 (Striping with double distributed parity):
  • Similar to RAID 5, but with two parity calculations
  • Higher redundancy, can survive two drive failures

Example: Imagine RAID 5 as a team of 4 writers working on a book. Three write content, while one summarizes (parity). If one writer is lost, their work can be reconstructed from the summary and other writers' work.