RAID hard disk drive configurations explained. How to balance cost, reliability, read/write speed, data recovery, and storage capacity.
RAID (Redundant Array of Inexpensive Disks) systems virtually join multiple hard disk drives (HDD) to form one logical hard disk. The computer sees that one logical hard disk and not the individual hard disks. There are different ways to join the hard disks, each with their own advantages and disadvantages.
For simplicity, hard disks of the same size are usually used. Many RAID systems will treat all disks as having the capacity of the smallest disk. Some proprietary RAID systems, such as Drobo, can better use disks of different capacities.
The RAID system can be implemented in hardware, or in software as an operating system device driver. Examples of hardware solutions include Adaptec (PCI disk controller card) and Asus (PC motherboard). Hard disk manufacturers (Seagate, Maxtor, etc) and server manufacturers (IBM, Dell, HP, Fujitsu, etc) are a source of RAID disk enclosures.
RAID 0
RAID 0 joins 2 or more hard disks together. “Striping” is used so that one block of data is split into a few sub-blocks. These sub-blocks are evenly spread across each disk in the RAID array. This improves file read and write times because data can be read or written to all disks at the same time.
Advantages
- Makes full use of available disk capacity. Storage capacity = n x s, where n is the number of disks, and s is the capacity of one hard disk.
- The faster file reads and writes compared to a single disk.
- Minimum 2 hard disks.
Disadvantage
- No redundancy. Failure of any single hard disk will disrupt the entire RAID array.
Applications
- Disk-to-disk backup. A RAID 0 array will reduce the backup time (window), and can be used as a high speed cache for a tape backup system. It is unlikely that both the main disks and the backup RAID array will fail at the same time.
- Writing large amounts of streaming data, such as the output from a high speed document scanner. If the data is lost, the documents can be scanned again.
RAID 1
RAID 1 is also called disk mirroring. Two disks are always used. Data is written to both hard disks, so each disk is a duplicate (mirror) of the other. When reading data, half the data can be read from one disk and half from the other. This speeds up file reads.
Advantages
- Can tolerate loss of one hard disk.
- File reads are faster compared to a single disk.
- Easier to recover data from crashed hard disks as there is no striping or parity data to consider.
Disadvantages
- File writes are the same speed as a single disk.
- Limited to 2 hard disks only.
- Storage capacity is the same as one hard disk.
Applications
- Boot disk for operating system.
- General purpose low-capacity disk for a good balance of reliability and read/write speed.
RAID 5
RAID 5 is like RAID 0, but with additional parity data stored. This allows the array to tolerate the loss of any single hard disk.
Advantages
- Higher read speed compared to one hard disk.
- Relatively high capacity. Storage capacity = (n-1) x s, or the same as all hard disks minus one.
- Can tolerate loss of one hard disk, but not two.
Disadvantages
- Lower write speed compared to RAID 0 of equal available capacity.
- Requires minimum of 3 hard disks.
Applications
- File server.
- General purpose high-capacity disk for a good reliability and high read speed, if the percentage of writes is low.
RAID 10
This is a combination of RAID 1 and 0. The disks are striped to create a larger capacity disk, and are mirrored. It’s also called RAID 1+0 or 0+1, depending on whether the disks are mirrored first or striped first, when the array is created.
Advantages
- Higher speed reads and writes compared to one hard disk.
- Higher capacity than RAID 1.
- Can tolerate at least one hard disk failure, possibly more depending on which hard disks fail.
Disadvantages
- Low usable capacity. Storage capacity = n / 2 x s, or half of the total hard disk capacity.
- Requires minimum of 4 hard disks.
Applications
- High performance database server, or other medium-capacity applications where reads and writes need to be fast.
The Best RAID Configuration
A RAID 1 array is a good general-purpose configuration for home and small office use, especially now that affordable 1-terabyte hard disks are available.
Choosing the right RAID level for high capacity (multi-terabyte) and/or high performance (read and/or write) mission-critical business system requires detailed information about the planned system (possibly based on the system it will be replacing):
- Number and size (bytes) of reads and writes per second (peak or busy-hour).
- Initial storage capacity and estimated annual capacity growth.
RAID arrays are often implemented as NAS drive enclosures. The Zero Assumption Software website has more details on the different RAID levels. There are specifics on reading and write speeds, RAID 6, and hot spares. Adaptec has a page on improving RAID write speeds. The BAARF website has links to detailed articles on RAID.