In: Computer Science
Consider the following tables depicting variants of an important data distribution technique used in the RAID technology to improve disk performance and answer the following questions (a & b ).
Table 1:
Disk 1 |
Disk 2 |
Disk 3 |
Disk 4 |
File 1, bit 1 |
File 1, bit 2 |
File 1, bit 3 |
File 1, bit 4 |
File 1, bit 4 |
File 1, bit 5 |
File 1, bit 6 |
File 1, bit 7 |
File 2, bit 1 |
File 2, bit 2 |
File 2, bit 3 |
File 2, bit 4 |
File 2, bit 4 |
File 2, bit 5 |
File 2, bit 6 |
File 2, bit 7 |
Table 2:
Disk 1 |
Disk 2 |
Disk 3 |
Disk 4 |
|
File 1, block 1 |
File 1, block 2 |
File 1, block 3 |
File 1, block 4 |
|
File 1, block 4 |
File 1, block 5 |
File 1, block 6 |
File 1, block 7 |
|
File 2, block 1 |
File 2, block 2 |
File 2, block 3 |
File 2, block 4 |
|
File 2, block 4 |
File 2, block 5 |
File 2, block 6 |
File 2, block 7 |
a) What is the technique known as and how it improves the disk performance?
b) What are the variants of the technique (identified in i) are known as? Explain how these variants are different from each other?
Answer:
(a) The technique is called RAID (Redundant Array of Independent Disks). This is a technique to cope with the risk of loss of data and lots of downtime, by improving disk usage. It is based in interleaving/striping disks and using them as one storage unit. It increases the disk transfer rate by a factor of N by splitting files across N different ydisks. Instead of saving all the data from a given file on a single disk, it is spread across many such disks. Now the heads of each of the N disks can search data independently , the speed of transfer is increased by nearly N times. The total transfer rate increases directly proportional to the number of disks. In this technique there is always maintained a backup or duplicate copy of each disk, become if anyone of the disks get damaged then all of the other disks get affected and all data is lost. Thus duplicating the disks also makes the storage system much more reliable.
-------------------------------------------------------------------------------------------------------------------------------------
(b)
Table 1 depicts RAID 2 level, called Stripping with ECC(Error Correction Code) , where disk striping takes place at bit level. That is each bit of the file data is spread across all the disks with a Hamming code parity bit.
Table 2 depicts RAID 4 level, called Independent data disks with shared parity disk. Here stripping takes place at block level. The file data is generally divided into blocks of 64KBytes .
The main difference is in the size of unit of data which is evenly spread across the disks. While the later provides better performance the former is more reliable. In both the levels, parity bits/blocks are used to reconstruct the data in case any of the disks crashes/fails.
------------------------------------------------------------------------------------------------------------------------------------------