How should RAID configuration be determined?

• Published: 30 Jan 2008

Q: What determines the choice of RAID configuration?

About the author Storage expert Kurtis Lindemann is the director of Technical Services for RoundTower Technologies, a storage service provider based in Cincinnati. Listen to the rest of Kurtis' answers on assessing a customer's storage capacity by downloading our storage capacity podcast.

There are three criteria to consider here: performance, reliability/availability and cost. If raw performance is the only concern your customer has and your array supports RAID 1 and 0, then that coupled with 15,000 RPM drives is the way you want to go. However, expect the price tag to be high for this solution. For normal workloads, RAID 5 with 10,000 RPM drives should be sufficient.

Just make sure that you have put in enough spindles to handle the customer's workload, and validate with vendor sizing tools to ensure there are no gotchas after implementation. If you're taking the "cheap disk" approach (that means SATA), then you absolutely want a solution that can handle RAID 6, or something RAID-6-like with a dual-parity scheme to protect from double-disk failures. SATA drives are less reliable than FC drives and are also typically quite a bit bigger. This means that rebuild times will be significantly longer for a 750GB drive than it is for a 146GB drive in a similarly sized RAID group. Since rebuild times are longer, it means you are susceptible to a parity error or a second drive failure that can bring a system down. RAID 6 protects your customer from this issue. The trade-off is that there is additional overhead from the second parity calculation, so you want to take this into account when sizing the solution.