Raid 5 Performance

It is needless to say that Raid 5 is one of the most widely used raid level. It consists of minimum 3 drives where both the data and parity is striped across all of them. Raid 5 performance is actually an overall estimation of how fast is its reading/writing, its storage efficiency and even rebuilding ability.

Raid 5 performance is heavily dependant on its reading/writing ability. Its write performance is terribly bad. When there are multiple small writes, it finds it difficult to cope up with it because of parity overhead. As the parity has to been updated in each write operation, the writes are really slow. In terms of reads, raid 5 performance is quite good. But then raid 0 reads faster as it does not have to omit the parity information of every stripe.

In terms of storage efficiency, it is quite impressive. The formula being:

• Storage efficiency = (Total number of disks-1)/ (Total number of disks).

This means if we take a 3 disk array, the storage space comes to 67%. So, greater the number of hard disks in an array, greater is the storage efficiency.

Raid 5 performance also depends on how fast the rebuilding procedure is taking place. When a disk gets corrupted, using the data stored in the other disks, the data in the corrupted disk is generated. Luckily, due to parallel processing, the rebuilding time for a 3 disk array is the same for a 20 disk raid array. Just the complexity in calculating the XOR for 19 disks in 20 disk array could be slightly greater than that for a 3 disk array.

During the array rebuilding process, if another disk gets corrupted then the entire array fails. This array failure is estimated as, Array failure= Chance that the disk fails * the total number of disks. It is clear from the expression that the greater the number of hard disks, greater is the probability of array failure. Thus, to increase Raid 5 performance, the number of hard disks should be the least. Ideally, raid 5 hard disk below 14 is acceptable.

Mean time between failures is the measure of the reliability of hard disks. The higher the value, the greater is the time period between two subsequent disk failures. Mean time between failures (MTBF) is reciprocal to the probability of the array failure rating. If the probability of failure increases, the MTBF decreases that is highly undesirable!