5.4.1 Raid

Learn how RAID (Redundant Array of Independent Disks) works within systems that have multiple drives.

RAID basically is a system that combines multiple storage devices under the management of one storage controller card and displays it as a single storage system. It improves system performance and decreases data redundancy.

Implementations

Back in the day, you would have to purchase and install a RAID controller card, which has its own processor and memory to create and manage the disk arrays. Nowadays, processors are so powerful that they themselves create and manage the disk arrays without taking a performance hit. However, for the most flexibility regarding configuration, RAID controllers would be the best way to go. There's also an OS method in which the processor uses RAID features that are integrated within the OS itself.

RAID 0: Disk Striping

Disk striping (a.k.a. RAID Level 0) takes data that needs to be stored and splits it into pieces, writing them across multiple disks. This results in blazingly fast read/write performance using the dedicated RAID controllers.

PROBLEM: If one of the drives fail, the data stored on both of the drives becomes corrupt. That's why the zero in RAID 0 means that there's zero redundancy and zero fault tolerance. Each additional drive creates another failure point, increasing the odds that someday the array will crash.

RAID 1: Disk Mirroring

Disk mirroring provides redundancy by making both disks copied reflections of each other, all storing the same data. Info that needs to be written to the disk is written on all drives at the same time. This is good because if one of the drives fail, the others pick up and the user doesn't notice anything at all. However, this means that all read/write requests pass through the RAID controller. If the RAID controller that's managing the mirrored drive array fails for any reason, all data would be lost.

RAID 1: Disk Duplexing

It's basically the same thing as Disk Mirroring except every drive gets its own RAID controller. Should one of the RAID controllers fail, the other working ones will pick up and manage the disks that were under the failed controller.

RAID 5: Striping w/ Parity

Parity is a calculated value that's used to rebuild data in case of a failure. You can store it on a parity drive. However, the parity value is usually striped across the entire array of drives, stored in a reserved partition of the drive.

RAID 5 requires three or more drives to implement. It's best practice for the drives to be of the same size and speed for peak performance.

In the case of any disks failure, the parity information can be used to rebuild the data that was stored in the lost disk.

RAID 10 (1 + 0): Mirror Striping

Uses both mirroring and striping. The best RAID technology.

RAID 10 requires four or more drives to implement. RAID 10 systems can only use half of the total raw capacity of the drives due to mirroring, but it provides a read speed that is twice that of RAID 5.

#V

#Aplus