August 6, 2008
Sean "Obsidian" Potter
Colin "Rhettigan" Dean
http://www.highpoint-tech.com
Forums
1 2 3 4 5 6 7
This site will look much better in a browser that supports web standards, but it is accessible to any browser or Internet device.

To look at the performance of the RocketRAID 3510, I'm going to test the drives in RAID 0, RAID 1, and RAID 5. For RAID 0 and 1, I'm only using two drives, so I'm able to compare results against that of the RocketRAID 3120. In RAID 5, I'm using four drives. The advantage of the hardware RAID controller over software RAID is primarily CPU usage. For machines used for high performance computing, software RAID is rarely an option, as it can utilize large amounts of CPU time depending on disk usage. For this reason, I'm going to disregard software RAID performance in this review.
hdparm stands for Hard Drive Parameters, and adjusts just that. Specifically, hdparm has a wide set of tools for ATA drives, such as adjusting the DMA. For other drives, the options become much more limited, as they rely more on hardware settings than software. Using the '-t' flag, hdparm returns the buffered disk reads, or essentially the average read speed. Higher is better.

dd takes an input, output, and various sizes and writes to disk. In our case, we always use /dev/zero, since /dev/zero continuously outputs 0 and a fixed rate (much faster than a hard drive can read / write). I'll be copying a little over 4GB to drive to test write performance. Let's look at the output of testing dd with the drives in RAID 5 (Higher is better):
# dd if=/dev/zero of=/mnt/raid/test bs=1024 count=4000000 4.1 GB copied, 45.8526s seconds, 89.3 MB/s
