RAID

Seite 1

RAID 0 (stripping)

A stripe set breaks data into units and stores the units across a series of disks by reading and writing to all disks simultaneouslty. RAID 0 uses two or more disks. Stripping:

Provides an increase in performance

Does not provide fault tolerance. A failure of one disk in the set means all data is lost

Requires a minimum of two disks

Has no overhead because all disk space is available for storing data

This is the fastest of all RAID types. However, it does not provide fault tolerance

RAID 1 (mirroring)

A mirrored volume stores data to two duplicate disks simultaneously. If one disk fails, data is present on the other disk, and the system switches immediately from the failed disk to the functioning disk. RAID 1 uses two disks. Mirroring:

Provides fault tolerance for a single disk failure

Does not increase performance

Requires two disks

Has a 50% overhead. Data is written twice, meaning that half of the disk space is used to store the second copy of the data

RAID 1 is the most expensive fault tolerant system

RAID 5 (stripping with distributed parity)

A RAID 5 volume combines disk striping across multiple disks with parity for data redundancy. Parity information is stored on each disk. If a single disk fails, its data can be recovered using the parity information stored on the remaining disks. RAID 5:

Provides fault tolerance for a single disk failure

Provides an increase in performance for read operations. Write operations are slower with RAID 5 than with other RAID configurations because of the time required to compute and write the parity information

Requires a minimum of three disks

Has an overhead of one disk in the set for parity information: (1 / n - 1)

A set with 3 disks has 33% overhead

A set with 4 disks has 25% overhead

A set with 5 disks has 20% overhead

RAID 10 (stripe of mirrors)

A RAID 10 volume stripes data across mirrored pairs and across multiple disks for data redundancy. If a single disk fails, its data can be recovered using the mirrored information stored on the remaining disks. If two disks in the same mirrored pair fail, all data will be lost because there is no redundancy in the striped sets. RAID 10 uses a max of four disks. RAID 10:

Provides fault tolerance for a single disk failure

Provides redundancy and performance

Is the best option for I/O-intensive applications like database, email, web servers, or any other system requiring high disk performance

Uses 50% of the total raw capacity of the drives is due to the mirroring

Requires a minimum of four disks

Hardware

Hardware RAID uses a special controller card that includes a RAID processor. Hardware RAID is the most expensive method bur provides much better performance and is more reliable than other methods

Software

Software RAID uses a driver and the system CPU for controlling RAID operations. This is the slowest form of RAID

Some RAID controller cards support RAID configuration, but without the onboard RAID processor. These solutions are classified as software RAID (sometimes called fake RAID) even though you install a controller card to provide RAID capabilities

Many motherboards include built-in (onboard) support for RAID. RAID implemented in this way is typically software/driver RAID

Operating System

Operating system RAID uses RAID features within the operating system. Like software RAID, the system CPU is used for RAID operations, but performance is typically better than software RAID because of integration with the operating system