Page 9: The advantages of RAID: Tests

We carried out a set of tests on our Athlon XP 2800+ system equipped with the Highpoint controller and the twin 120GB Seagate ATA-5 drives. Using three different hard disk benchmarks, plus some timed file copies we tried to get a picture of just how much of a real-world performance boost a hardware or software RAID 0 array will give as compared to a single drive. We also tested a RAID 1 configuration to see where it fits in terms of speed.
Futuremark's brand new benchmarking program, PCMark 2004, includes 4 separate hard drive performance tests. These use Intel's Rankdisk application to record an isolated simulation of various common Windows disk access events. The performance of the drives in these events is then rated based on the amount of data they were able to transfer per second.
PCMark 2004 HDD Benchmarks - Results
Physical Drive Size XP Startup Application Loading File Copying General HDD Usage
A) No RAID, Single HDD 120GB 7.168 6.536 22.778 5.085
B) Hardware RAID 0 240GB 10.281 7.050 32.104 6.588
Software RAID 0 240GB 10.535 6.984 29.426 6.713
Hardware RAID 1 120GB 8.396 5.391 15.353 5.364
Units: MB/s MB/s MB/s MB/s
The results of the tests speak like this, for the "XP Startup" tests RAID of any sort seems to be an advantage. When we consider "Application Loading," there isn't much of advantage for RAID 0, though you can see a performance handicap imposed by RAID 1 begin to appear. The "File Copying" tests definitely show the advantage of having a RAID 0 setup. Lastly, the "General HDD Usage" tests show surprisingly little difference between a software and hardware RAID 0 array, at least on a fast computer.
Sandra is designed to test the theoretical power of a complete system and individual components. The numbers taken though are again, purely theoretical and may not represent real world performance.
Sisoft Sandra 2004 - Benchmark Results
Avg. Access Time Buffered Sequential Random
Tests: Time: Read: Write: Read: Write: Read: Write:
Single HDD -Standard IDE 7 ms 87 50 41 42 8 9
Hardware RAID 0 6 ms 82 52 80 55 9 11
Software RAID 0 6 ms 82 52 81 55 9 11
Hardware RAID 1 3 ms 89 28 42 27 15 9
Units: milliseconds MB/s MB/s MB/s MB/s MB/s MB/s
You can see the advantage in terms of reading and writing speed that RAID 0 gives, as well as the slow writing performance of RAID 1 quite easily in this benchmark. Still no difference in performance between hardware RAID 0 and software RAID 0.