which one of these results would be considered the most accurate
and why such a large range
and why do people that run 2.55 with a non changeable size seem to get better results
and is there a normal setting to run it on considering no one usually mentions what they run it at
sorry for lots of questions but now im curious
Your disk array will perform differently depending on what you ask it to do. Changing your benchmark settings will not affect how fast your disks run. The trick is finding which setting most closely represents the way YOU will use YOUR disks. Then your benchmark results can be useful with array performance tuning.
Messing about with benchmark settings to get the highest possible sequential reads, then tuning to further enhance those sequential reads will likely hurt your performance if the settings you end up optimising for do not represent the way you use your disks. If, for example, you use your array for large media files, optimising for the large stripe sizes would be the way to go. If however you use your array for OS use then smaller stripe sizes and higher IOPS would probably be your target. Optimising for sequential reads then using the disks for OS will result in less than optimal real-world speeds, despite your chosen benchmark results saying you are running as fast as possible.
I'm afraid there is no 'one best setting' for the benchmarks - they all have their use, hence the reason why the settings are user-adjustable.
Edited by the_beast - 7/16/09 at 1:56am