RAID and Subsystem Architectures RAID Configurations
RAID is the ability to read or write data across multiple spindles. It uses internal memory in each disk to avoid waiting whilst the heads are positioned to the correct track and the disk rotates so the required sector is accessible by the head.
RAID 0, 4, 5 and 6 all use block level striping An individual disk would contain the 512 Bytes typically associated with a block command stored on one sector. This is often referred to as INDEPENDENT ACCESS as an individual block could be read from or written to an individual disk without accessing the other drives in the stripe set. If an individual block is written to however, the parity block(s) will still need to be updated for RAID 4, 5 and 6. This could create a hot spot with fixed parity solutions and that is why most vendors chose to implement distributed parity solutions, unless they could guarantee that the entire stripe would be written to each time. Slightly higher write overhead when compared to RAID 2 or 3
RAID 2 and 3 use byte level striping An individual disk would contain individual Bytes of the 512 Bytes typically associated with a block command. This is often referred to as PARALLEL ACCESS as an individual block would have to be read from or written to all drives in the stripe set. This means that all drives in the stripe set are always accessed and therefore there is no benefit of using distributed parity to minimise any hot spots.
Typically used for large sequential access by individual applications or users.