<== Return to Hardware Section
RAID
Redundant Array
of Independent (or Inexpensive) Disks
*** Should you use RAID? No !! ( see RAID - not such a Clever Idea for the Home - there are actually tons of articles about the problems with RAID)
*** instead - invest your money in a Fast drive, with a good backup drive. Don't mirror - instead run overnight backups.
There are two main types of RAID that home users will buy - both are problematic :
RAID 0
- if one drive goes bad - they both go bad. It is 6 times more likely
to lose data than regular single-drive systems
RAID 1 - even though it was made for the sole purpose of protecting data loss - RAID 1 is actually slightly more likely to lose data than regular single-drive systems, because it adds complexity to the book-keeping of the data Reads and Writes, and is more susceptible to errors. Invest in a fast drive with a backup, instead. Also, there's the perceived notion by the users that the mirrored drive is the same as a backup - NOT TRUE. The same feature that provides the protection can also be the user's downfall. RAID 1 maintains a faithful copy on the second disk of everything that's on the first. Warts and all. The mirror includes mistakes, files irrecoverably deleted, viruses, spyware, and if the first drive gets trashed through improper data being placed on it - then the backup drive is equally trashed !! Users tend to forget that RAID 1 does not protect against errors, it protects only against one disk going faulty.
*** see also http://www.acnc.com/04_00.html for an excellent overview
RAID is a configuration of 2 or more hard drives, that uses various algorithms that are based on two concepts:
Here we discuss RAID for use with PC's and Servers - and the technical aspects of RAID (all the various RAID Levels). RAID has been in use with servers for many years, but recently has become popular with PC's as well, as the capacity demands have increased, and RAID prices have come down. Some people call these "RAID Arrays" but this is a redundant use of the term, since the acronym has the word "array" in it.
The five most Common RAID Levels
RAID
0 (RAID Striping) - overrated, do NOT use it !!! Gamers and Video enthusiasts are excited about the
supposedly reckless but blazingly fast RAID 0 (RAID Striping), which allows your PC to
access two drives simultaneously, but offers no redundancy or protection
whatsoever. This effectively doubles the speed of your drives access
rate. Most say RAID 0 should not even be called
"RAID". So RAID 0 is a tradeoff. BUT it is actually
more reckless than it is fast. (see this artcile,
Single Drive vs
RAID 0 ).
Why you should not use RAID 0
it is only as good as its weakest Link - all drives in a RAID 0 array should be identical. If not, you can still run RAID 0, but each drive will be treated as if it is the same size as the smallest drive in the array. In addition, and all drives will be accessed at the same speed as the slowest drive
RAID 0 substantially increases your risk of disk failures, and in fact it "often" causes disk failures. In addition, both disks fail simultaneously !!
like the concept of dual processors where few vendors have written
software to take advantage of them . . . few
software vendors have written software that takes advantage of striped
drives - therefore in the vast majority of cases, there is little or no real
speed benefit !!
RAID 1 (RAID Mirroring) - unlike RAID 0, this is "real RAID". It uses a "redundant" configuration, where if one drive fails your system will not lose ANY data. Copies and maintains an identical image of data from one drive to a second drive. If one drive fails, the disk array management software directs all applications to the surviving drive as it contains a complete copy of the data in the other drive. This RAID configuration provides data protection and increases fault tolerance to the entire system. Use two new drives or use an existing drive and a new drive for this setup. The new drive must be of the same size or larger than the existing drive.
RAID 0+1 (a RAID 1 Mirror of two RAID 0 Stripes) - some people believe this is the best RAID, because it takes care of the deficiencies of RAID 0. It offers the benefits of both RAID 0 and RAID 1, and requires no parity data or parity checking. Implemented as a mirrored array whose segments are RAID 0 arrays. Has the same fault tolerance as RAID level 5. But . . .
it is VERY EXPENSIVE because it requires 4 disks !!!
like RAID 0, few software vendors have written software that takes advantage of striped drives - therefore in the vast majority of cases, there is little or no real speed benefit !!
RAID 5 (the "Business" RAID) - stripes both data and parity information across three or more hard disk drives. Among the advantages of RAID 5 configuration include better HDD performance, fault tolerance, and higher storage capacity. The RAID 5 configuration is best suited for transaction processing, relational database applications, enterprise resource planning, and other business systems. Use a minimum of three identical hard disk drives for this setup.
All RAID Levels
There are a number of different ways of doing this, and each method has been assigned a RAID Level, such as RAID Level 0, RAID Level 1, etc. These levels are often called simply RAID-0, RAID-1, etc.
The Hard Drive Bottleneck - RAID to the Rescue
In 1987, UC Berkeley researchers David Patterson, Garth Gibson, and Randy Katz warned the world of an impending I/O crisis because server hard drives simply could not keep up with all the advancements in the speed of the other components (CPU, RAM, etc). They realized the large mainframe hard drives were the bottleneck of the entire system (actually - to this day - hard drives are still the bottleneck). The team assessed the various advantages/disadvantages of these huge "mega-drives":
At that time, large disk drives for mainframes were incredibly expensive ($35,000 was not unusual), but they noticed that the common PC drives had become fast and cheap, especially SCSI drives.
Idea 1 - Combine Multiple Drives - they came up with a key concept, to lash 75 PC disk drives together, and created a controller which combined their storage into one , big, virtual drive. They then assessed the advantages/disadvantages of this solution and compared them with the current mainframe drive :
- Multiple Drives Advantages - same high capacity, lower cost, faster speed (I/O was 12x faster !!)
- Multiple Drives Disadvantages - poor reliability
The solution seemed great - except the reliability was poor, because now there were 72 cheap assembly-line devices to worry about breaking down, instead of 1 expensive high-quality device.
Idea 2 - Redundant Drives - to tackle this problem, they came up with the second key concept in RAID technology. They suggested the use of extra "check" disks, containing redundant information that could be used to completely recover data in the event of a disk failure. Once a failed disk was replaced, either by a human operator or by electronic switching, data would be reconstructed onto it automatically.
Their team then proposed this solution in a paper, and coined the acronym, RAID. The paper was titled "A Case for Redundant Arrays of Inexpensive Disks (RAID)."
Within a couple of years, Intel-based products like the Compaq Systempro (released in 1990) made RAID an expected ingredient in every midrange and high-end server.
The RAID Acronym Fiasco - it is a rare occurrence when vendors change the meaning of a standardized acronym - but in this case, they did. The word "Inexpensive" was true for the first test model of the combined 72 PC drives. But as you can expect, vendors don't make $$$ on inexpensive components - so the RAID arrays (initially SCSI, then both SCSI & IDE were available) on the market were actually very pricey. The Vendors temporarily decided that the "I" stood for "Independent" instead. Now everyone wonders why there are two meanings for the acronym.
RAID with PC's
RAID arrays were expensive, and only used with mainframes and servers for many years. But recently they have come down in price, and although still a niche market - they are coming on quickly and are available by default with new PC's - or as upgrades to existing PC's. Businesses need the added reliability for their power user workstations, and home users need the added speed for games, video, etc.
RAID Controller - Integrated MOBO vs Add-On Card - most new motherboards now have RAID support built-in (integrated RAID). For existing systems, you can buy a RAID controller card (the most popular is Promise Technologies).
Matched Drives - while it is not mandatory to have identical hard disks with RAID - it is very highly recommended !! Actually, if you can - get identical drives - Capacity, Make, and model.
The 3 Types of Hard Drives - make sure that the RAID controller you use will work with the type of hard disk you have. Most will work with all three primary drive types (IDE, SATA, SCSI), but there are some that do not.
NOTE on SATA Drives (Serial ATA Drives) - the 40-pin parallel ATA IDE and EIDE drive interface has been the standard for years, with SCSI running far back in second place. In 2002, Serial ATA (SATA) was introduced as the next step in ATA technology. SATA provides greater scalability, simpler installation, thinner cabling, and faster performance (up to 1.5 Gb/s). SATA maintains backward compatibility with the Parallel ATA software drivers, and is planned for a speed increase of up to 6Gbps in coming years.
Be aware that when you install a serial drive, by default it will show up in Windows as a SCSI drive !! I am not sure why, but it may be that Windows developers did not envision SATA drives, and therefore do not have that text in their list of all possible devices.
*** see also www.serialata.org
RAID 0 + 1 will give you the best of both worlds.
You do not need to have your operating system on the RAID drives. It is quite possible to have Windows on a 40 GB drive called C and have two 250 GB hard disks setup in RAID 0 as your drive D for video clips
If going for SCSI RAID note that RAID cards that have dedicated hardware on the card to control the RAID data handling are superior to the software based RAID controller cards.
When buying a motherboard ensure that it supports not just IDE RAID but Serial ATA RAID as well. When buying a server motherboard you can get SCSI RAID on board.
Check that your RAID controller has support both for RAID 0 and RAID 1 AND that it has support for RAID 0 + 1; you never know when you may want to add more disks to your RAID.
If you want the drive containing Windows to be in a RAID array, setup the RAID prior to installing Windows.
RAID 0 (Striping) - Don't do It !!!!!!
Striped Disk Array without Fault Tolerance
RAID 0 substantially increases your risk of disk failures, and in fact it "often" causes disk failures. In addition, both disks fail simultaneously !! Worst of all, almost no software exists that uses the simultaneous data access in an efficient manner - so although it does offer much faster data access - it has very little real speed impact on the applications because they continue to operate on one piece of data at a time.
RAID 0 is an oddity . . . the "0" means that there is zero RAID (no redundancy). So, oddly - RAID 0 is not RAID !!
RAID 0 simply stripes data across the drives - so if you have two 100 GB drives, for example, you have one, big, 200 GB virtual drive. Since there is no backup of the data, RAID 0 - or RAID at Level 1 is the only Level that does not offer data redundancy.
STR simply does not significantly impact performance of typical desktop applications.
There are certain uncommon situations where RAID 0 can significantly improve system performance. For example, editing of large audio or video files is sometimes limited by the maximum sequential transfer rate of the hard drives, but it is far more common for the processor to be limiting factor.
So why not just use the two drives separately then ?? Well, RAID 0 designed systems use hard drive controllers that can Read and Write data faster if they alternate between drives. The bytes can actually be written to the two drives simultaneously, but keep in mind that a single CPU machine can only work with one byte at a time. Nevertheless, the CPU is much faster than hard drives, so it supplies the drive buffers with as much data as they can possibly work with. So bottom line - you get faster data transfers.
BUT if one of the drives fails, all drives go down.
Disadvantages
RAID 1 - Mirroring
Disadvantages
