Storage Management

Activate your FREE membership today  |  Log-in

  • Visit other TechTarget ANZ sites: 
Posted
Nov 19, 2007
 |  By:  TechTarget ANZ Staff

RAID FAQ Part One

Bookmark and Share

RAID is old yet still developing, useful yet mysterious. Our FAQ series aims to clear up some of the fog that surrounds the data storage schemes that RAID comprises.

Q: What are the different RAID levels, their characteristics and when to use each level?

A: First and foremost, RAID can be used to enhance availability, but it also can be used to enhance performance for IO, for throughput, things like that.

So, real quickly RAID 0 -- data spread across multiple drives, gives you performance but no availability; RAID 1 -- data mirroring. Mirroring data across two or more disk drives gives you availability and can also enhance read operation; RAID 01 or RAID 10 -- combine striping and mirroring or mirroring and striping for a compromise of good performance and good availability; RAID 3 -- striping with dedicated parity. Good for large sequential bandwidth operations but not as good for concurrent access. That's where RAID 5 steps in, which stripes the data across multiple drives with rotating parity. This gives you a nice balance of good read performance and reduced cost, because you don't need to have as many drives for mirroring. It's a compromise between performance and availability at the expense of writes, because you have parity updates. Now, you are starting to see dual-parity schemes, such as RAID 6 -- multiple drives striped together with at least two parity drives.

Q: RAID is now 20 years old, is it still applicable today as it was in the past?

A: Yes it absolutely is. RAID is becoming more common and transparent to the point that it doesn't have to be hyped and it loses its appeal, so to speak. Even though disk drives are more reliable, we are seeing more large-capacity drives being deployed and in different applications. RAID may not be needed for some scenarios, but we're seeing RAID trickle down into the SMB [small and medium-sised business], as well as further out into the enterprise. So, whether it's a software or hardware internal RAID controller or an external RAID array, RAID continues to proliferate.

Also, there are new forms of RAID emerging, including distributed parity, local and remote parity, and dual-parity schemes.

Q: With a shift toward solid-state disk and semiconductor memory-based storage, does that diminish the need for RAID to boost performance?

A: No, I'd say that RAID is still necessary. RAID and solid state complement each other. There's a notion that RAID is just for disk drives, and that's not the case. One solid-state vendor, Texas Memory Systems, recently released a solid-state storage system that combines traditional RAM cache with multiple flash-based modules, with the flash modules actually configured with RAID 5 for availability and read performance using the read cache to offset the RAID 5 penalty. This is a great example of a hybrid systems, as well as hybrid drives, complementing each other with the RAID for availability and performance.

Even some general purpose x86 class servers have been shipping for several years now with mirrored RAID -- RAID 1 for availability and redundancy purposes. It's a good fit across different technologies, and it's not just for disk drives anymore.

NEXT: More on RAID


TechTarget ANZ sites: SearchCIO.com.au | SearchNetworking.com.au | SearchSecurity.com.au | SearchStorage.com.au | SearchVoIP.com.au

WF Online community sites: ElectricalSolutions | ElectronicsOnline | FoodProcessing | InMotionOnline | LabOnline | ProcessOnline | RadioComms | SafetySolutions | SustainabilityMatters | Voice&Data

Copyright © 2010 Westwick-Farrow Pty Ltd. All rights reserved.
About Us | Contact Us | TechTarget