Earlier we used to have a single Disk which involved a lot of time delay and low performance in terms of fetching data and responding. This gave a rise to thought of how we could increase performance of disk while enhancing reliability. Then, came concept of parallel performance and concurrent Disk access.
Redundant Arrays of Inexpensive Disks (RAID) was a new concept introduced which could do the simultaneous access of files. This concept was later called Redundant Arrays of Independent Disks. Using RAID, storage could be optimized with better performance and more reliability.
The main ideology backing performance enhancement is: Data Stripping. This involves Distributing Data among various disks and accessing all disks in a simultaneous manner to retrieve data. The main concept of increasing performance in case of multiple disks involves distributing data into several disks instead of keeping it in only one disk, using concept of parallel data access. To achieve this, there are 2 concepts that can be followed. They are bit-level data striping and block-level data striping.
- Bit-level data striping can be done to enhance disk performance. This is a very molecular concept. We have to take ith bit of each byte of data. Then, place it in ith disk. For example, if we take 8 disks, one for each bit of a byte (of data), data transfer would have an 8-fold increase. The reading and writing operations would involve use of all disks. This would immensely increase speed of operation.
- If we store ith block of data on ith disk, we call it Block-level Striping. For example, if we use 8 disks, handling of 8 simultaneous block requests can be done. If multi-block data request is present, then, in that case, increase in transfer rate would be enhanced by 8 times. A generalized view of this is that using n number of disks, performance could be increased by n times.
In bit-level striping and block-level striping, data access is paralleled and speed is increased. We use a number of disks, which enhances speed and performance. However, when there is failure in any of disks, entire operation would be affected. This system hence proves itself very less reliable.