How much would adding a third drive to the raid array help? Just wondering if it would be worth it, $55.
For the WEI score: it wouldn't change anything. For the 4k random read: I don't know exactly, 0.8 MB/s maybe. It still wouldn't break 1 MB/s. For desktop usage you probably wouldn't notice the difference. The downside would be that it doubled the chance of failure.
Its about as consistent as my child's bathroom habits (he's 2). So whether its a score, a measure, w/e, its useless b/c its not consistent. It pains me to see anyone use this to gauge anything b/c of that.
I agree and disagree. The WEI score is useless to actually get any detailed understanding of how your machine performs. But it is not because it is inconsistent. It is an amalgam of different benchmark results, and this makes it inconsistent. The real problem is that it is opaque. If you knew how the score was calculated you could based on your score figure out how it actually did in the benchmarks. Microsoft know this so they can have windows make decisions based on the WEI score. Like they enable or disable different hard drive performance tweaks based on the WEI score.
For benchmarking hard drives things like HD Tune and HD Tach is better but they have problems as well. They are designed to show the benchmark performance difference between hard drives, not the performance difference you actually experience. Sequential read speed is the one in which you get the biggest differences so it is what they focus on.
Things like Crystal Disk Mark and AS SSD benchmark add 4k random read and write, which is better, but they are doing the same thing again in focusing on the benchmarks which show the biggest difference amongst SSDs. It is not as if real desktop IO is limited to either sequential or 4k random. And aside from AS SSDs 64thred test, they generally only test with a queue depth of 1.
Real desktop usage is an unholy mix of different sized random and sequential IOs in different sizes, with queue depths of ~4 and sometimes up to around 10, and non of the benchamrks out there really test that. Sure there is IOMeter, but it is more server oriented. There are also thing like PCMark, but it had similar disadvantages as WEI in being opaque.
Ok, that is enough ranting form me.
