Fully Automated RAID Level Selection Techniques for Disk Arrays Page 5


Want the latest storage insights?

Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure

Summary And Conclusions

The focus of this article has been on a general variety of methods for automatically selecting RAID levels, running the gamut from ones that consider each store in isolation and make irrevocable decisions to ones that consider all workload interactions and can undo any decision. The simpler tagging schemes are similar to accepted knowledge and to the back-of-the-envelope calculations that system designers currently rely upon. However, they are highly dependent on particular combinations of devices and workloads and involve hand-picking the right values for many constants, which makes them suitable only for limited combinations of workloads and devices.

Integrating automated RAID level selection into a store-to-device assignment algorithm leads to much better results, and, as a result, the benefits of a fully-adaptive scheme outweigh its additional costs in terms of computation time and complexity.

Finally, for future work, implications should be explored that provide reliability guarantees in addition to performance. Fully-adaptive schemes would be suitable for this, albeit at the cost of increased running times. Thus, the automated selection of components of different cost for each individual LU within the arrays (i.e. deciding between big/slow and small/fast disk drives according to the workload being mapped onto them) extends the automated decisions to additional parameters, such as LU stripe size and disks used in an LU.

[The preceding article is based on material provided by ongoing research at the Storage and Content Distribution Department, Hewlett-Packard Laboratories, 1501 Page Mill Road, Palo Alto, CA 94304 on RAID level selections. Additional information for this article was also based on material contained in the following white paper:
"Selecting RAID levels for disk arrays" by Eric Anderson, Ram Swaminathan, Alistair Veitch, Guillermo A. Alvarez and John Wilkes at: http://www.hpl.hp.com/personal/John_Wilkes/papers/FAST2002-raid-level-selection.pdf]

John Vacca is an information technology consultant and internationally known author based in Pomeroy, Ohio. Since 1982, John has authored 39 books and more than 485 articles in the areas of advanced storage, computer security and aerospace technology. John was also a configuration management specialist, computer specialist, and the computer security official for NASA's space station program (Freedom) and the International Space Station Program, from 1988 until his early retirement from NASA in 1995. John was also one of the security consultants for the MGM movie titled : "AntiTrust," which was released on January 12, 2001. John can be reached on the Internet at jvacca@hti.net.

Submit a Comment


People are discussing this article with 0 comment(s)