Tuesday, May 15, 2012

What is RAID?

If you are in the computing business where the data security and availability are crucial at all times, chances are you might have already encountered or even worked with RAID. What is RAID? It is not a living thing but it is has become quite popular and seriously vital in many organizations with multiple users across different locations or even home users accessing multiple pictures at a time. RAID is an acronym for Redundant Array of Independent Disks or sometimes referred to as Redundant Array for Inexpensive Disks. The latter has been coined mostly to illustrate capability for generally improving computing performance and data security by using multiple low-cost disks.
In order to get down to the different efficiency pros and cons of various RAID levels, it is more proper to lay down some clarity to many RAID terminologies. Striping is how data are distributed and written into the various disks that will be set up into array either by byte-level or bit-level. Mirroring is the process by which data written into the other disk is copied onto the other disks simultaneously, creating a redundant replica of the other. Parity on the other hand is use alongside with striping and it is computational procedure that allows the array in determining what information has to be stored into what disk.

Standard Levels

Depending on the priority of the user, there are RAID levels that can be set to give more emphasis on some organizational and business decisions crucial to the operation such as increased computing performance, data security, data availability, and data storage capacity.
This array is mostly used by users who pay more for high performance computing. The adage that says “two heads are better than one” holds true for this set-up. It allows but offers no mirroring or redundancy in the process. The increased performance is made possible with the way data are stored and split into several disks in the array. This works wonders for those on the look out for speeding up the read and write computing functionality.
This set-up offers no striping but allows mirroring in the process. As data are identically written on various disks in the array, there is a very slim chance of losing any data even if the other disks fail in the process. However, performance is a little bit compromised compared to RAID0 array as the write process to a single disk may take a bit more time compared to having them split into several disks.
It is implemented by splitting data at the bit level and spreading it over designated data disks and redundancy disks. The redundant bits are calculated using Hamming codes, a form of error correcting code (ECC).
This array is set-up where striping is on a byte-level and with dedicated parity disk. This means that every file is split up amongst the striped drives equally, byte-by-byte, and this is what makes it exceptional because it allows reading of every byte simultaneously across different drives.
This array set-ups a block-level striping and setting aside a disk for dedicated parity). It is similar to RAID 5 except that it confines all parity data to a single disk which can negatively impact on the speed performance. Nonetheless, the error detection is achieved through dedicated parity which is stored in a separate, single disk unit.
This is set-up allows the bit level striping and distributed parity calculation. This is has been the more popular set-up among the earlier set-ups mentioned as it offers the benefits of RAID3 and RAID4 but without the bottleneck experience caused by a dedicated parity computation in a single disk.
This set-up offers the bit-level striping and double distributed parity calculation. This also means more costly on disk arrays as you would have to make available double quantity of the disks. Across the board it offers reasonable performance but slightly lower than what RAID5 allows. It can expensive, too.

Nested (Hybrid) RAID

There are also arrays that employ the combinations of standard RAID levels by attaching both the standard RAID level numbers before and after a “+”, such as 0+1.
RAID 0+1: This refers to striping of data in mirrored sets. Although this has become more complex for ordinary users, this allows fault tolerance and improved performance.
RAID 1+0: This refers to mirroring of data in a striped set.
RAID 5+1: This refers to the set-up with mirrored striped set and with distributed parity.
The above descriptions and set-up explanations are not all the time applicable to low-end users and may not even be comprehensible to some high-end computing professionals. Nevertheless, knowing that there are possible disk solutions to your computing needs is always a useful piece of information.

No comments:

Post a Comment