RAID<sup>n</sup> to increase reliability


At last month's Comdex show, Land-5 Corp. unveiled technology that it claims will increase reliability compared to traditional RAID levels, or architectures. Designed to protect disk arrays against multiple drive failures, RAIDn is a set of software algorithms that support parity checksum calculations and can handle any number of drive failures.

Land-5 demonstrated the technology embedded in the company's StoragePod network-attached storage (NAS) systems, which are based on a Linux kernel. For now, RAIDn works only with Linux, although the C code could be ported to other operating systems in the future. In addition to embedding the technology in its NAS servers, Land-5 plans to license RAIDn to other vendors such as RAID controller manufacturers.

In the NAS configuration, RAIDn buffer management code sits between the file system and the SCSI block device architecture. The code could also be implemented in firmware on RAID controllers, working in conjunction with XOR engines.

RAIDn provides the same reliability as RAID 1+5, with fewer drives and at lower cost.
Click here to enlarge image

Land-5 officials argue that most conventional RAID arrays can protect against only one drive failure or against three drive failures in very costly mirroring-plus-parity configurations. RAIDn allows administrators to protect against n-1 drive failures, where "n" indicates the total number of drives in the array.

In addition to reliability improvements, company officials claim potential cost savings and, compared to some RAID levels, improved performance, particularly in read operations. For example, Land-5 CTO Kris Land says RAIDn performance is equivalent to RAID 5 performance and is faster than RAID 1+5 or RAID 5+5 configurations. He also claims RAIDn can provide equivalent reliability of RAID 1 mirroring with about half the number of drives.

For more information, browse the company's Website at www.land-5.com.

This article was originally published on December 01, 2000