For our power consumption testing, we have the drive connected to the system as a secondary drive. To record the wattage, we are now utilizing a Quarch Technology Programmable Power Module. It allows us to accurately measure power consumption over time and is flexible enough to allow us to test any SSD that comes our way.
When enterprise class storage is deployed, not only one or two drives are set out in a deployment, hundreds to thousands are, therefore, when booting a NAS or SAN it is important to ensure the power sources will be able to provide enough power to the systems. First we look at how much power the device needs during startup. Next we look at idle power consumption and workload averages at a QD of 256. This helps us gauge IOPS per Watt.
Our power analysis may change as time goes on, but for now we are looking at just a few metrics with the main goal of measuring our results against the manufacturer’s ratings.
Overall, we can see that power consumption is well regulated to the tune of up to an average of just under 19W when under load. These results are marginally better than that of the 1.6TB Intel P3608 (with both 800GB volumes in RAID 0) we’ve reviewed last year, however, when looking at write power consumption we can see it clearly uses much less power. The same goes for startup and idle consumption, again, it is very well controlled compared to the Intel HHHL card.
Now, looking at the IOPS per Watt under workloads we can see that during the write tests the SN100 still boasts an advantage, however, during read workloads the Intel takes the lead. We can see that especially clear when comparing at the Web Server workload. One thing to keep in mind still is that, most likely due to a compatibility issue with our system and the SN100 we were not able to achieve the rated 4KB read IOPS, therefore, results may differ in your own test systems when qualifying this drive.
REPORT ANALYSIS AND FINAL THOUGHTS
The HGST Ultrastar SN100 gave us a stellar performance throughout benchmarking. When we first put it through its paces in our system we weren’t able to achieve the rated 4KB performance, however, after a UEFI update we were able to without a problem. Comparing our synthetic workloads, it gave the Intel DC P3608 a run for its money, especially during the server workloads. The Intel had the advantage during the Web Server workload, but the SN100 beat it out during three of the four workloads.
Furthermore, reviewing the power consumption results, the Intel DC P3608 flaunts its advantage during the read centric tests by delivering much more IOPS per Watt than it. HGST Ultrastar SN100, on the other hand proved itself to be marginally more efficient during mixed and write heavy workloads compared to the Intel DC P3608. Overall, we are very impressed with the HGST Ultrastar SN100’s performance.
HGST has developed a highly competitive SSD suitable for the most demanding scale-out database workloads. From both a device design and management standpoint the HGST Ultrastar SN100 is a solid package all around. It has, all the enterprise features one could ask for, such as power-loss protection to end-to-end datapath protection and RAID like parity to ensure your data is safe. Paring together both PMC’s robust NVMe controller and a plethora of Toshiba eMLC flash, the Ultrastar SN100 is capable of delivering sequential speeds upward of 3GB/s read and 1.6GB/s write as well as random 4KB speeds up to 743K IOPS read and 140K IOPS write. Best of all, this is all within a 2.5″ or HHHL AIC form factor with capacities up to 3.2TB.
At its current prices as of our finding (though not including bulk discount), this SSD looks to be very enticing given the performance we have seen here today. Sure, it doesn’t have the endurance of say an Intel DC P3700, but for what it is, the SN100 is a reliable product that comes in at a very good value.