RAID Controllers

RAID, or Redundant Array of Independent Disks, is a method of grouping hard disks together in order to store the same data in different places. RAID uses as few as two or as many as 32 hard disks, and is created by using a controller either in the form of an expansion card (CERC/PERC) or embedded on the motherboard (ROMB or RAID On Motherboard).

By implementing RAID, you can achieve:

Performance: By striping data across more than one disk, your data retrieval speeds increase.
Protection: With mirroring, data from one disk is copied to a second disk. One disk can fail, and no data is lost. With parity, data is divided across multiple disks. If one disk fails, data is recreated from the surviving disks.
Combination of performance and protection: Using striped mirroring, data is divided across a set of mirrored disks.

	CERC ATA100/4CH	PERC3/Di	PERC3/SC	PERC3/DCL	PERC3/DC	PERC3/QC	PERC4/Di	PERC4/SC
PCI Slot Speed	32-bit/33 MHz	N/A-ROMB	32-bit/33 MHz	64-bit/66 MHz	64-bit/66 MHz	64-bit/66 MHz	N/A-ROMB	64-bit/66 MHz

SCSI Channels Ultra3/160	N/A	2	1	2	2	4	2	1

ATA Channels	4	N/A	N/A	N/A	N/A	N/A	N/A	N/A

Physical Connectors	4 internal	2 internal	1 internal	2 internal 2 external	2 internal 2 external	2 internal 4 external	2 internal, 1 external	1 internal, 1 external

Cache	16MB	128MB	32MB	64MB	128MB	128MB	128MB	64MB

Backup Battery	No	Yes	No	No	Yes	Yes	Yes	No

On-board Processor	i960 100 MHz	i960 100 MHz	i960 100 MHz	i960 100 MHz	i960 100 MHz	i960 100 MHz	i303 100MHz	Intel 80302

RAID Levels	0, 1, 5, 10	0, 1, 5, 10	0, 1, 5, 10, 50	0, 1, 5, 10, 50	0, 1, 5, 10, 50	0, 1, 5, 10, 50	0, 1, 5, 10	0, 1, 5, 10, 50

Bandwidth	100MB/sec	160MB/sec	160MB/sec	160MB/sec	160MB/sec	160MB/sec	320MB/sec	320MB/sec

RAID Level	Features	Advantages	Things to Consider
RAID-0: Striped Disk Array	With striping, data is divided across multiple disks (or spindles). Data is broken down into blocks, and each block is written to a separate disk. The workload is reduced for all disks, helping to accelerate data delivery times. RAID-0 helps make disks more responsive, especially for e-mail, database, and Internet applications. RAID-0 requires at least two hard disks to implement.	Greatly improved performance by spreading the I/O load across many disks. Easy to implement.	RAID-0 offers NO data protection. Should not be used for critical data.

RAID-1: Mirroring	RAID-1 is achieved through what is called disk mirroring, and is done to ensure data reliability. The same data is copied and stored on different disks. If one disk fails, the data is available somewhere else in the array and can be easily restored. Mirroring not only creates redundant data for high availability, but keeps critical applications running as well.	Increased performance for reading data, with the same write performance as a single disk. 100 percent data redundancy means there is no data rebuild in case of a disk failure - simply copy the data from the remaining disk in the pair.	Inefficient use of disk capacity - subsequently has the highest overhead of all RAID types (100 percent).

RAID-5: Striping with Parity	RAID -5 maintains data redundancy through a technique called parity checking. As data is striped across multiple disks, parity bit data is included and also distributed among all the disks in the array. Parity is used to maintain data integrity and rebuild lost data in case of a disk failure. If one disk in the array fails, the missing data can be reconstructed from the remaining parity bits on the surviving disks. A RAID-5 configuration requires at least three disks.	Most efficient use of disk capacity of all the redundant RAID configurations. Maintains good read-write performance.	Disk failure can impact throughput rates. Reconstructing information after a disk failure takes longer than with a mirrored configuration.

RAID-10: Striping of Mirrored Arrays	RAID-10 is a combination of RAID-1 and RAID-0. This configuration requires at least four disks, and offers the best performance, protection, and capacity of all the RAID levels. RAID-10 consists of pairs of mirrored disks whose data is striped across the entire array. In many cases, RAID-10 can sustain multiple simultaneous disk failures making it less susceptible to downtime. It has the highest probability of no data loss.	Same redundancy as RAID-1 (mirroring). Best choice for data protection.	Can be expensive, due to the mirrored disk array.

RAID-50: Striping of Distributed Parity Arrays	RAID-50 is a combination of RAID-5 and RAID-0. This configuration stripes data with parity across each RAID-5 subset of disks. Each RAID-5 subset requires three hard disks. RAID-50 provides a higher degree of fault tolerance since one disk per subset may fail without data loss. Since the parity bits are distributed across the RAID-5 subsets, data rebuild speed is increased.	Greater degree of fault tolerance. Potential for faster data read rates.	Disk failure impacts throughput. Reconstruction of information after a disk failure takes longer than with a mirrored solution.