BBS | Home | Mainboards | CPU | Networking | Memory | Video | Tech Advice
What's new | Cartoons | email me | About the author | Links | Advertising Info

A Word about RAID

RAID is a common insecticide or bug killer and in a way the analogy holds also for another RAID which stands for one of the most interesting backup systems, the Redundant Array of Inexpensive Disks, originally described in 1987. The term now has been re-coined to mean Redundant Array of Independent Disks (or Devices) and what it circumscribes is a way of treating several independent hard drives as one unified device with combined (higher) capacity and multiplied transfer rates (higher speed). Another advantage of the RAID concept is the avoiding of data loss owing to corrupted files and thatís were the analogy to the bug killer comes in :) .

The current versions of RAID include several levels or configurations called 0, 1, 0+1, 2, 3, 4 and 5, each of which has its unique advantages and features.

Level 0
Level 0 is a later addition to the original paper on RAID and often it is not even considered real RAID because there is no redundancy involved in the setup, that is, the entire set of data is split into several chunks which are stored independent from each other. For example in fast digital image analysis where storage of data always lags behind the acquisition system and constitutes a bottleneck for the possible temporal resolution of the system (I am talking about 60 to 80 MB of data /sec) the images can be split into several frames which are treated individually and stored individually, including the necessary input/output functions. For example using an array of 4 HDs will cut the storing time not quite in four but close enough. Upon replay, the data files have to be spliced again to reconsitute the original image but again, since the individual fragments come from separate drives the possible transfer rates can be multiplied roughly by the number of devices. And, of course the complete file/folder can be transferred to another system to enable other systems to view them. Other applications include video editing and 3D graphics.

Level 1:
Level 1, which is also called mirroring of the drive, essentially uses 2 different HDs to store identical sets of data. In other words, if one HD is corrupted or shows any other failure, it can be exchanged for another drive onto which the second drive is copied which still holds an exact copy of the first drive. The advantage is clear, there is only a negligible chance for data loss because of disk failure, however the disadvantage is that twice the amount of disk space is needed (actually two separate disks) and there is no increase in speed because of division of files into data chunks.

Level 0+1
Level 0+1 is, as the name indicates, a combination of the two above described levels and combines the capabilities of both levels. In other words, the data are striped among multiple drives to increase speed while at the same time mirroring is done for safety features. The simplicity and capabilities of this setup make it a widely used choice.

Levels 2 and 3
Levels 2 and 3 are by now pretty outdated and have been replaced by the other levels which offer better performance.

Level 4 Level 4 uses like Level 0 striping of the data onto different drives resulting in higher speed. Unlike Level 0+1 there is no mirroring involved, instead the data protection is achieved by adding parity. Parity is written to one dedicated disk in the setup as a way to store summary information of all data on the other disks within the array. Since no actual copy of all files is kept, the amount of space required for the protection is way less than what would be necessary for mirroring of the entire array. Still, the parity disk can be used to reconstruct all data on a disk that has been corrupted. That, unfortunately does not include severe mechanical damage of the drive, in which case data may be lost.

Level 5 Level 5 differs from Level 4 in that there is no dedicated parity disk in the array. Instead the parity is distributed, along with the data across the entire spectrum of disks within the array. This way, also the potential bottleneck of a single parity disk can be avoided. Also, because the individual disks are not synchronized, the array can carry out multiple read and writes at the same time which is an essential function for database management, transactions etc.
General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site!

All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.

BBS | Home | Mainboards | CPU | Networking | Memory | Video | Tech Advice