What Is RAID?

[contactTout name=mini]

Redundant Array of Independent Disks (RAID)

RAID is a data storage capability that combines multiple drives into a single logical unit and controls the way in which data is written to those drives to enhance data read/write performance, security (redundancy), or both. RAID storage adds various protections from potential data loss or drive failure.

RAID accomplishes disk protection in varying combinations of three factors:

Mirroring – Replication of data onto separate drives. Mirrored data is identical. Striping – Process of spreading portions of data across multiple drives so the data can be accessed faster by utilizing the speed of each drive, concurrently. Parity – An error-check and redundancy methodology, where a 1 or 0 is added to the end of a piece of code to indicate if the number of bits within that string of code is even or odd. The parity can be stored on the same drive as the data, but is usually stored on a separate drive. Parity helps with catching errors during write process and with recreating lost data using the remaining parity information from other drives to deduce what the missing data was. Parity mirroring is also an alternative to full mirroring to maximize available storage. Rather than duplicating the complete data set by mirrors, parity mirroring only duplicates the small bits of parity need to be mirrored onto multiple drives that would be needed to recreate the original data

Striping (RAID 0) for High Performance

RAID 0 is written across multiple disks in a round-robin fashion. There is no backup or redundancy, and thus no additional tasks to slow down the read/write speed. Advantage of RAID 0 is to achieve high performance by accumulating each individual hard disk performance. However, if any one hard disk gets defective, information stored in this RAID 0 will become invalid. RAID 0 has access to the maximum amount of drive space, but is vulnerable to a single disk failure rendering data on all other disks unusable. diagram-Raid0 Raid 1

Mirror (RAID 1) for High Security

RAID1 is known as Mirror-RAID. All original data from a disk(s) is replicated onto an additional disk(s) in real time to ensure continuous availability. In a RAID 1 system with two hard disks, the data in one hard disk will be an exact copy as the data in the other hard disk. The RAID system will use half the available disks/space as mirrors, giving you half the usable capacity of available storage. Since data is being written twice, there is also a write performance delay, though read speeds are largely unaffected. Failure of a single disk will cause the RAID controller to enter into degrade mode, but will still be functional and all data available/recoverable. The host controller can still read/write data to the RAID but it will be slower. Once the dead drive is swapped out, the RAID will rebuild automatically. RAID 3 diagram-Raid5 diagram-Raid6

Parity Protection (RAID 3 and RAID 5 and RAID 6)

RAID controllers generate parity block via the XOR engine. A parity block is small piece of data used to recover data that might be missing or corrupted. In RAID 3 mode, all parity data will be stored on a dedicated “parity-only” drive. As long as the parity drive does not fail, data from any of the other drives can be recovered and restored.   In RAID 5 mode, parity block will be spread over all of the different hard drives in an alternating fashion. Since the parity data is distributed, any single drive can fail and the data can be recovered and restored.   In RAID 6, the parity block will be distributed just like RAID 5 except there will be two copies of all parity blocks, each copy-pair being on different drives. This means any two drives can fail, and all data can still be recovered and restored.

 Storage Comparisons

The RAID controller will also make use of the same size disk space in each hard disk under RAID 3 / RAID 5 condition. Under RAID 6, due to 2X the parity, there is a decrease in available storage in exchange for double redundancy. In a 6-drive set up, RAID 3/5 has 17% of storage dedicated to parity, and under RAID 6 has 33% of storage as redundant data. Failure in a hard disk will cause the RAID controller to enter into degraded mode. Host controller still could rear/write data thru the RAID normally without knowing any defects. Users have to replace the defective hard disk. The RAID controller will then enter into online-auto-rebuild mode automatically.

Stripe + Mirror (RAID 10) for High Performance and High Security

RAID subsystem could be configured to support Stripe and Mirror at the same time, i.e. RAID 10. Take four hard drives in RAID 10 as an example. Hard drive 0 and hard drive 1 could act as Mirror 1. Hard drive 2 and hard drive 3 act as Mirror 1 too. The RAID controller then configures these two Mirrors as Stripe. At least two drives (either one in each Mirror) can be allowed to fail without any impact to RAID data access. If two drives from the same Mirror are failed, the RAID data is not accessible and becomes invalid. diagram-Raid10 Raid clone


Clone’s action is similar to RAID 1. However, all of the hard disks will be the mirrors. For example, in a four hard drives Clone environment, data in each hard drive will be the same. This mode is useful especially when users would like to copy data from a hard drive to several hard drives at the same time. The number of allowed failed drives is the total number of drives in the RAID minus one.

Concatenating (Large)

This mode is also named “Large”. In this mode, the RAID controller will concatenate all of the hard drives into a single hard drive with larger capacity. For example, if three 500GB hard disks are connected to the RAID subsystem in Large Mode, user will get a single hard disk with capacity of 1,500GB. If any one hard disk gets defective, information stored in this LARGE RAID will become invalid. RAID Large










No – Parity data is randomly mirrored





Yes – data and parity

Yes – data and parity






Yes – Double


Performance Penalty X = spindles * drive speed





0.5X (Can be as fast as RAID 0 if using 2X the amount of drives and capacity as RAID 0)

How Many Drives Can Fail Before Data is Lost





50% of drives

Disk Requirement

At least 2

At least 2, must have even amount

At least 3

At least 4

At least 4


Fastest Most drive space

 Full duplication.  Accomplishes redundancy via parity without needing to use 50% of drives to create mirrors.

2X more secure than RAID 5 Virtually no downtime when 1 drives fails Most popular RAID

Most secure and most performance. Best throughput of any other RAID set up, other than RAID 0. Can handle more drive intensive operations


Most prone to complete data loss. Absolutely no back up

Half of drive capacity is used as duplication

Not good for large data. When 1 drive fails, performance to recreate drive puts large strain on remaining drives

 Slightly less storage than RAID 5 due to 2X parity

More drives mean more expensive.

Half of drive capacity is used as duplication