Steady-state Iometer
The benchmarks thus far have shown performance in the context of what's expected on a new drive. Behind-the-scenes garbage collection is able to keep the performance at near-new levels at all times, helped by the fact that there's lots of free space on each drive.
But what happens if we deliberately push the drives into states they will likely be in when used for months at a time? The premise here is to see how performance differs once the SSDs' garbage collection system is working overtime, which happens when there's no free space on the Windows-accessible area.
So we've filled each drive to its absolute limit by writing out a capacity-encompassing Iometer iobw.tst file. Filled to the brim, which takes a long, long time on the Crucial BX200, any subsequent action requires the SSDs to employ all their technologies in order to process the benchmark as quickly as possible. The way in which the over-provisioning works is key to ensuring solid performance.
This is a worst-case scenario for an SSD, of course, and real-world use isn't likely to be so pernicious to performance, but it's instructive in understanding the real quality of an SSD's controller.
Once full, we set the drive to run the Iometer 4K random write test at queue depth of 32 for an hour. We then log what happens in the last five minutes when the drive really is in a beaten-up state.
This first test shows the average IOPS during this last five-minute period. The results are a far cry away from the kinds of figures you see printed on the box; those are derived when it's in a fresh state.
Speed is one facet; the other is how consistent the drive is under heavy duress. It's worth knowing just how variable the results are from one moment to the next. The best drives exhibit very little standard deviation; leading to smooth, predictable performance. The Crucial does reasonably well in this regard, actually.
As a final metric for a well-used drive, we can look at both performance and consistency by dividing one by the other. A great drive will score highly here because it'll possess huge performance and very little standard deviation. The Crucial BX200's relatively slow performance and average consistency combine to put it at the bottom. This is the gold standard by which any drive can be judged.
