Results: Latency And Performance Consistency
Random Performance Over Time
My saturation test consists of writing to each drive for 12 hours using 4 KB blocks with 32 outstanding commands. But first I secure erase each drive. Then, I apply the write load, illustrating average IOPS for each minute (except for the last 20 minutes, where I zoom in and show you one-second average increments).
The write saturation test has always been more about characterizing drive behavior than pure performance. In that way, it's necessary to push client SSDs harder at times. Once they start sweating, we get a better idea of what might be going on under the hood.
This chart comes from The SSD 730 Series Review: Intel Is Back With Its Own Controller. The 100% write (in pink), 50% write (in green), and 30% write (in blue) workloads are tightly grouped. More important, as the workload gets progressively more read-biased, speed improves. Writes are the limiting factor, so as we turn the dial up on those, distinct performance bands emerge with the 100% write workload on the bottom.
As you're about to see, most drives don't look as neat and orderly as Intel's exception SSD 730. But you get the idea.
After hammering on the SSD340 for a while, we start seeing some unusual activity. The drive creates the tightly-grouped bands seen in the chart above, with little separation between minimum and maximum performance. We just don't get much of an improvement as the workload shift from write-only (most demanding) to 50% writes to 30% writes. The SSD340 is value-oriented, so it's understandable that Transcend doesn't sacrifice usable capacity for more over-provisioning, which would have helped improve degraded performance. Instead, you get 7% spare area, and that's probably going to hurt the SSD340 through our testing.
