AMCC 9500S User Manual

Page of 103
Understanding RAID Levels and Concepts
www.3ware.com 
9
Understanding RAID Levels and Concepts
                                      
3ware RAID controllers use RAID (Redundant Array of Inexpensive Disks) 
to increase your storage system’s performance and provide fault tolerance 
(protection against data loss). 
This section organizes information about RAID concepts and configuration 
levels into the following topics:
“RAID Concepts” on page 1
“Available RAID Configurations” on page 1
“Determining What RAID Level to Use” on page 1
RAID Concepts
The following concepts are important to understand when working with a 
RAID controller:
Arrays and Units
. In the storage industry, the term “array” is used to 
describe two or more disk drives that appear to the operating system as a 
single unit. When working with a 3ware RAID controller, “unit” is the 
term used to refer to an array of disks that is configured and managed 
through the 3ware software. Single-disk units can also be configured in 
the 3ware software.
Mirroring
. Mirrored arrays (RAID 1) write data to paired drives 
simultaneously. If one drive fails, the data is preserved on the paired 
drive. Mirroring provides data protection through redundancy. In 
addition, mirroring using a 3ware RAID controller provides improved 
performance because 3ware’s TwinStor technology reads from both 
drives simultaneously. 
Striping
. Striping across disks allows data to be written and accessed on 
more than one drive, at the same time. Striping combines each drive’s 
capacity into one large volume. Striped disk arrays (RAID 0) achieve 
highest transfer rates and performance at the expense of fault tolerance. 
Distributed Parity
. Parity works in combination with striping on RAID 5, 
RAID 6, and RAID 50. Parity information is written to each of the striped 
drives, in rotation. Should a failure occur, the data on the failed drive can 
be reconstructed from the data on the other drives. 
Hot Swap
. The process of exchanging a drive without having to shut 
down the system. This is useful when you need to exchange a defective 
drive in a redundant array.