Intelligent RAID 6 Theory Overview and Implementation
RAID 5 systems are commonly deployed for data protection in most business environments. However, RAID 5 systems only tolerate a single drive failure, and the probability of
encountering latent defects of drives approaches 100 percent as disk capacity and array width increase. To address this issue, RAID 6 systems will soon be widely adopted because
RAID 6 systems protect critical user data even when two drives or more fail concurrently. The Mean Time to Data Loss (MTDL) of RAID 6 systems is greater than RAID 5 systems;
therefore, RAID 6 systems provide much better data protection than RAID 5. The RAID 6 theory is based on Maximum Distance Separable (MDS) coding, as well as Galois Field (GF)
mathematics, which is introduced briefly in this paper. In addition, the P and Q parity computation and data recovery schemes are presented to help readers transition to RAID 6
systems rapidly. Lastly, the RAID 6 acceleration features of the Intel® IOP333 I/O processor are discussed. The IOP333 processor offloads the host CPU from intensive RAID 6
computations performed on a byte-by-byte basis in the data-path, allowing implementation of RAID 6 without compromising performance.
- File Name/Size:
- 30812202.pdf
- 2073418 bytes
- Download From:
-
FTP Server
-
Web Server (Available for byte serving)
|
