To read more on our test methodology visit How We Test Enterprise SSDs, which explains how to interpret our charts. The most crucial step to assuring accurate and repeatable tests starts with a solid preconditioning methodology, which is covered on page three. We cover 128KB Sequential performance measurements on page five, explain latency metrics on page seven and explain QoS testing and the QoS domino effect on page nine.
We test sequential throughput with multiple threads (up to 8) to unlock higher performance, but it is noteworthy that several threads issuing sequential workloads in tandem can lend some randomness to the workload.
With that in mind, the 16-SSD SanDisk array provides a blistering 13GB/s of throughput during the measurement window. The 16-SSD Intel array follows with 10.4 GB/s. The 8-SSD SanDisk and Intel arrays, which are spread among two containers to maximize throughput, provide better scaling and offer 9 and 8 GB/s of throughput, respectively.
The Intel and SanDisk arrays continue to jockey for position between the varying read or write workloads. The 16-SSD Intel array takes the lead at 256 OIO with 13.6 GB/s. The 16-SSD SanDisk array offers the best performance from 1 to 32 OIO, but at 256 OIO it drops to a lower performance threshold. The performance 256 OIO drops below the 8-SSD SanDisk array's 10 GB/s of throughput.
A result with that type of disparity can be indicative of thermal throttling, but we monitored the temperature of various SSDs during the test (via vendor utilities) and found them to operate within a safe thermal envelope in all workloads.