| Raid for dummies 1 The term .RAID became popular because of some articles about making systems fault tolerant. The name originally stood for Redundant Array of Inexpensive Disks. Marketing executives changed Inexpensive to Independent so they could charge more.
Note: There is not really a period before .RAID (RAID), the period keeps machine translators from changing the word into something about battles or wars.
The disk drive is the most mechanicaly challenged part of every computer. If you run it long enough or hard enough, it will fail. When this happens, people get upset. Bosses scream and shout. Orders don't get shipped. It is not a nice thing.
.RAID keeps the system running when a drive fails. Repairs can be scheduled for Saturday. It is much less stressful to be around the .IS manager or the Boss.
.RAID is a productivity tool. It is used to avoid lost productivity due to a drive failure. In 1993 an estimated Five Billion dollars ($5,000,000,000.00) of productivity was lost because of computer down-time. This cost continues to rise. Fault tolerance and redundancy are an increasingly important economic choice.
.RAID levels 1, 5 and 10 (0+1) are in common use. Level 1 (Disk Mirroring) is the most widely used. It keeps a complete, on-line backup copy of a drive instantly available.
Level 5 (Cheaper) and 10 (Faster) makes a bunch of disks look like one big disk. This is only necessary if you need a really big volume or file, or you have Array envy.
.RAID can be done with software, hardware or a combination of both. Software solutions use your main computer to do the .RAID work. While it is doing this it cannot do other things as quickly. Pages don't load as fast. Perceptive Solutions offers software and software-hardware hybrids (DataShadow family) for cost challanged applications.
Hardware solutions can be internal or external. External sub-systems are more costly and have more parts to fail. Perceptive Solutions specializes in internal solutions for OEMs and VARS. Our higher performance controller (Intelli-Cache) uses its own processor and memory. The controller takes care of the disk sub-system and improves performance with caching. Your expensive server can run at full speed, games and pictures load faster. If someone is working, they can continue working even if a drive fails.
Centari Computers excells in support for OEMs and VARs needing cost effective, high performance fault tolerant systems.
RAID-0 SPAN, usually 2 or 4 drives. Not fault tolerant, used to make large volumes and to explain level 5 and 10.
RAID-1 MIRROR, uses 2 drives in place of 1. Usually provides lowest cost and best performance. Most widely used (counting commercial and industrial).
RAID-(0+1) MIRRORED SPAN, usually 4 or 8 drives. A RAID-0 span mirrored to a redundant set of drives. Cost is higher than RAID-5, but performance is better, especially with a failed drive. Use in place of RAID-1 for really big arrays.
RAID-5 SPAN with PARITY, usually 3 or 5 drives. Lower cost upgrade to RAID-0 than (0+1). Slower because of parity, especially with a failed drive.
RAID-10 Another name for RAID-(0+1). Easier to say than zero plus one.
SCSI SCSI drives still dominate very large requirements and high reliability requirements. SCSI has been around longer and is more mature. There is significant focus by SCSI vendors on error handling and reliability. Larger capacity drives are only available in SCSI. Attractive for large server and very high reliability requirements at any level.
IDE As larger IDE (E-IDE, UDMA) drives become available, they become attractive for low cost .RAID use. IDE was developed for the cost sensitive consumer market and such drives do not always reflect the maturity, reliability and error handling of SCSI. Wide variation is found in actual speed, reliability, and error handling quality of the drives.
|
