TR DriveBench 2.0 — More disk-intensive multitasking
As much as we like DriveBench 1.0's individual workloads, the traces cover only slices of disk activity. Because we fire the recorded I/Os at the disks as fast as possible, solid-state drives also have no downtime during which to engage background garbage collection or other optimization algorithms. DriveBench 2.0 addresses both of those issues with a much larger trace that spans two weeks of typical desktop activity peppered with multitasking loads similar to those in DriveBench 1.0. We've also adjusted our testing methods to give solid-state drives enough idle time to tidy up after themselves. More details on DriveBench 2.0 are available on this page of our last major SSD round-up.
Instead of looking at a raw IOps rate, we're going to switch gears and explore service times—the amount of time it takes drives to complete an I/O request. We'll start with an overall mean service time before slicing and dicing the results.
Well, I wasn't expecting that. The old Samsung 830 Series has a quicker overall mean service time than the new 840 Pro. The difference amounts to only a tenth of a millisecond—less than the gap between the 840 Pro and the 840 Series—but it's enough to drop Samsung's new hotness to the middle of the pack. Updating the drive's firmware doesn't improve its performance, although that may be due to a limitation in the test. DriveBench 2.0 isn't capable of executing TRIM commands while playing back traces, and it's the handling of those commands that has changed in the 840 family's new firmware.
We can learn a little more about what's going on in DriveBench by splitting the mean service time between read and write requests.
The Samsung 840 Pro Series sticks to the middle of the pack in both metrics. Its mean read service time is reasonably close to the quickest drive of the bunch, at least, but its mean write service time is nearly double that of the leader (and of the 830 Series).
There are millions of I/O requests in this trace, so we can't easily graph service times to look at the variance. However, our analysis tools do report the standard deviation, which can give us a sense of how much service times vary from the mean.
The 840 Pro's read service times are nearly as consistent as those of the steadiest drive, but the difference is much larger with writes.
We can't easily graph all the service times recorded by DriveBench 2.0, but we can sort them. The graphs below plot the percentage of service times that fall below various thresholds. You can click the buttons below the graphs to see how the Samsung 840 Series compares to SSDs from other drive makers.
The write distribution plots are too closely matched to draw any real conclusions apart from the fact that the mechanical drive and Intel 320 Series aren't in the same league as the faster SSDs. The read results are more spread out, and they reveal that the Samsung 840 Series and 840 Pro have similar service time distributions beyond the 0.2-millisecond threshold.
Although the SandForce-based drives have higher percentages of service times under 0.1 ms, the 840 Pro pulls ahead as the threshold rises up to 0.6 ms. All the SSDs start to converge after that point.
As the distribution plots illustrate, service times over 100 milliseconds make up a tiny fraction of the overall results. Those extremely long service times have the potential to cause the sort of hitching that a user might notice, so we've graphed the individual percentages for each drive.
No problems here. The 840 Pro Series gets stuck on only a tiny fraction of the ~40 million I/Os in our two-week trace.