What makes you a geek?

[Concillian]

Here's one of the articles from there, take a close look at it before you keep touting that SSDs blow spindel drives out of the water.

SAS RAID vs. Mitron and Samsung controller SSDs

I can't find the other right now that was directly an Intel 80-M vs. the pair of Raptors, but as you can see from the above article, more often than not, the spindel drives had higher I/O than the SSDs when placed into a RAID configuration.

Did you read the article and realize that they're formatting the drives to ~1/10th their full capacity in order to make the rotational drives approach SSD performance? So you're arguing that we should not only use RAID, but also cripple the capacity in order to compete with SSDs? The full capacity formats don't compete with SSDs, except the 4 15kRPM drives in RAID0. So yes, they can compete if you like buying 4 15kRPM SAS drives and a SAS RAID controller.

Mostly this demonstrates exactly the points I've been trying to make. Look at the 4x 7200 RPM drives in RAID-0 (at full capacity format). They barely hang with a single 15kRPM drive at full capacity, and they don't hang with the SSDs at all. You can't RAID your way to lower latencies and RAIDing 7200 RPM or 10k RPM drives only gets you so far.

When seek times are very low (15k RPM 450GB drives formatted to 20 GB capacity) then, yes RAID-0 scales well. Because latency is reduced to the point where throughput has a much higher weight on overall performance. You can also see this in the "random access" IOmeter profile (web server). Here the SSDs dominate. Consequently, most users have a really large number of small files that they access on a regular basis (ever looked at the distribution of file sizes on your drive?). Actual performance for normal users is approximated with the Workstation profile, but in my experience, this still weights random small files too lightly. The delta in the small random profile (web server) is so large that even if you give it only a 5% weigting compared with workstation, it almost doubles the SSD scores. So even a small amount of small file random type of activity will be perceived as a benefitby an end user. I don't pretend to know an exact %, but in my experience, it's enough to be a noticeable win for SSD in actual perceived performance from the person sitting in front of the computer while doing actual things on the computer which is really difficult to quantify in a benchmark. It's like the single core / dual core transition a few years ago. Dual core gave real perceptible benefits that just weren't well captured in bencmarks.

So yes, if you're willing to de-stroke your drives to an abysmal capacity, then rotating storage can hang with SSDs if they're in a RAID. This article does a good job of demonstrating how much hinges on reducing latency, which is what I've been saying all along... latency is the dominant factor in end-user performance.

I do thank you for finally linking something, it was an interesting read.

I'll leave the thread with an interesting quote from Anand about how deceptive benchmarks are with SSDs:

[comments are based on a pretty old SSD that was sold with the Macbook Air]
The benchmarks for that drive didn’t really impress. Most application tests got a little slower and transfer speeds weren’t really any better. Application launch times and battery life both improved, the former by a significant amount. [...] It benchmarked faster than hard drive, but the numbers didn’t justify the cost. I pulled the drive out and sent it back after I was done with the review.

The next time I turned on my MacBook Air I thought it was broken. It took an eternity to boot and everything took forever to launch. Even though the benchmarks showed the SSD shaving off a few seconds of application launch time here and there, in the real world, it was very noticeable. The rule of thumb is that it takes about a 10% difference in performance for a user to notice. The application tests didn’t show a 10% difference in performance, but the application launch tests, those were showing 50% gains. It still wasn’t worth $1000, but it was worth a lot more than I originally thought.

It was the MacBook Air experience that made me understand one important point about SSDs: you don’t think they’re fast, until I take one away from you.

These are the exact kinds of comments people were making about dual-core processors. They were faster in benchmarks, but nothing to write home about... until you experienced one and then turned off one of the cores. There's more small file random reads than you think in a normal user's usage profile. Also consider that this is an experience with one of the first consumer SSDs, 2 generations older than current SSDs, which have made very significant advances.