RAID 5 explained: RAID 5 sees the distribution of parity information between all drives in the RAID set and is resilient against the failure of a single disk. So, how does RAID 5 share data in a 10 MB file between five drives?
When a LUN is created on a RAID group one of the settings used is the block size. This setting determines how much data is written to the drive at once and a common block size used in many environments is 64 KB. For example, in RAID 5 a 64 KB chunk of space is written to each drive in the RAID group to form a stripe. So, if we have five drives in the RAID set, a single stripe of data would contain 256 KB of data plus associated parity rebuild information.
In RAID 5, that parity information is distributed across all drives in the RAID set, so for each consecutive stripe across the RAID set, the parity information is on a different physical spindle. In our example of five drives, the first stripe would contain actual data on drives 1 to 4 and parity information on disk 5. The second stripe would contain data on disks 1, 2, 3 and 5 with parity information on disk 4, and so on.
In our example of a 10 MB file size, we would get an almost identical amount of data on each disk -- that is, 2 MB. The only difference in data capacity on each disk would be if the stripe count was not an exact multiple of five, in which case there would be a small number of stripes where the data was not split eq
This was first published in May 2011