Conclusion

The SATA SSD market is fairly mature, and in many respects our performance testing boils down to confirming that the bottleneck for a particular workload is the SATA link itself rather than the drive. For the most part, the Kingston DC500 SSDs check the necessary boxes, saturating the SATA interface for reads (random or sequential) and coming pretty close for sequential writes. Differentiating the DC500R and DC500M from other enterprise SATA SSDs requires digging a bit deeper.

The Kingston drives inherit a recurring problem we've noticed with Phison-based drives – sub-par sequential IO performance at QD1 – but they are competitive once queue depths climb a bit. Our tests of pure random writes turned out much better than expected for both Kingston drives, though the degree to which the DC500R beat its specifications was variable. Unfortunately, this good performance didn't carry over to the mixed read/write tests, where the Kingston drives merely matched the Samsung competition at best. The Aerospike Certification Tool showed that the Kingston drives both had much worse latency QoS than the Samsung SATA drives for a mixed workload that's a bit more write-heavy than the target market for these drives.

But these performance disparities are all dwarfed by the power efficiency gap between the Kingston DC500s and Samsung's SATA SSDs. Samsung's drives consistently use less power than the Kingston drives, and usually deliver equal or better performance. On some tests one or both of the Kingston drives match the efficiency of the Samsung entry-level NVMe SSD, but Samsung's SATA platform simply has unbeatable efficiency. It would take a fairly large pile of drives for this efficiency gap to seriously affect TCO including power and cooling costs, but it's something to watch out for, especially since the DC500 sometimes substantially exceeds its rated power draw when hit with a lot of writes.

The Phison S12DC controller used by the Kingston DC500s is a significant advance over their older SATA controllers and the move to 28nm undoubtedly helped keep power draw in check while adding the more robust error correction that modern SSDs need. But after hitting the SATA performance wall Samsung pivoted to reducing power for their SATA drives, and at the moment they're well ahead of everyone else on that score.

It's often hard to do any meaningful price comparison of enterprise SSDs, because many models are only or primarily sold in volume directly from the manufacturer to major customers. Those drives often show up at grey market resellers with pricing that is not at all indicative of the volume prices; spotty availability and the occasional surplus of a specific SKU cause major price distortions. This doesn't apply to the Kingston DC500 series, since they actually are intended for retail sales in individual quantities. The Samsung SATA drives in this review are also sold through distributors, so we can make a reasonable comparison. In the table below, we're using current prices from CDW, and we've included a few other relevant competing models with similar endurance ratings. All of these drives are likely to have volume pricing that's lower than these prices, but this should be an accurate picture of their relative positioning.

Enterprise/Datacenter SATA SSD Price Comparison
Unit Quantity from CDW, June 24, 2019
  Endurance
(DWPD)
480 GB 960 GB 1920 GB 3840 GB
Kingston DC500R 0.5 $104.99 (22¢/GB) $192.99 (20¢/GB) $364.99 (19¢/GB) $733.99 (19¢/GB)
Kingston DC500M 1.3 $125.99 (26¢/GB) $262.99 (27¢/GB) $406.99 (21¢/GB) $822.99 (21¢/GB)
Samsung 860 DCT 0.2   $174.99 (18¢/GB) $349.99 (18¢/GB) $699.99 (18¢/GB)
Samsung 883 DCT 0.8 $119.99 (25¢/GB) $219.99 (23¢/GB) $419.99 (22¢/GB) $799.99 (21¢/GB)
Intel
D3-S4510
1.4–2.1 $131.99 (27¢/GB) $215.99 (22¢/GB) $427.99 (22¢/GB) $824.99 (21¢/GB)
Micron 5200 PRO 1.3–2.5   $196.99 (21¢/GB) $373.99 (19¢/GB) $1221.99 (32¢/GB)
Micron 5200 ECO 0.6–1.1 $101.99 (21¢/GB) $187.99 (20¢/GB) $355.99 (19¢/GB) $651.99 (17¢/GB)
Seagate Nytro 1351 1.0 $118.99 (25¢/GB) $211.99 (22¢/GB) $394.99 (21¢/GB) $743.99 (19¢/GB)

Kingston's pricing for the DC500 SSDs is generally in line with the rest of the market. At the higher capacities, the premium for the DC500M over the DC500R is much smaller than we would expect given how much extra flash and DRAM the -M version includes for the same usable capacity. Strictly comparing the two DC500s, the -M seems to be offering a pretty good deal for the improved write performance and endurance. However, the Samsung 883 DCT offers similar throughput and better QoS for slightly lower prices than the DC500M, with the caveat that the Samsung drive has 40% lower rated write endurance. And it's important to keep in mind that the more write-heavy workloads where the DC500M stands out from the DC500R and the Samsung drives are also the workloads least likely to be run on a SATA SSD of any brand—that's NVMe territory now.

While we have not personally evaluated its performance, the Micron 5200 ECO looks pretty well positioned to compete against the DC500R: it's rated for similar write performance, but aside from the 7.68TB model it has twice the write endurance, and it's cheaper across the board.

What's Next?

Kingston has shared some of their plans for enterprise and datacenter SSDs going forward. The DC500 family is just two of several product lines they are launching this year. They haven't released specifics  about what other market segments they will be going after in the near future, but at the very least we expect an NVMe drive, probably an entry-level M.2 model based around the Phison E12DC or one of their later datacenter controllers. Since the datacenter drives take much longer to qualify than consumer products, we may not see a PCIe 4.0 datacenter drive before the end of the year even though consumer drives using the E16 controller will be hitting the shelves very soon.

Meanwhile for the successor to the DC500 family, Kingston is planning to update to Intel's 96L TLC, but that transition is probably at least a year away. To support this broadening of their enterprise/datacenter offerings Kingston has significantly increased the staff dedicated to these product lines, so we can expect a more regular cadence of updates—but still at enterprise product cycle pacing, which is not as fast-moving as the consumer SSD market has been lately.

Mixed I/O & NoSQL Database Performance
Comments Locked

28 Comments

View All Comments

  • Christopher003 - Sunday, November 24, 2019 - link

    I had an Agility 3 60gb, used for just over 2 years for my system, mom used now an additional over 2.5 years, however it was either starting to have issues, or the way mom was using caused it to "forget" things now and then.

    I fixed with a crucial mx100 or 200 (forget LOL) that still has over 90% life either way, the Agility 3 was "warning" though still showed as over 75% life left (christmas '18-19) .. def massive speed up by swapping to more modern as well as doing some cleaning for it..
  • Samus - Wednesday, June 26, 2019 - link

    I agree, I hated how they changed the internals without leaving any inclination of a change on the label.

    But the thing that doesn't stop me from recommending them: had anyone ever actually seen a Kingston drive fail?

    It seems their firmware and chip binning is excellent. The later of which is easy for a company that makes so many God damn USB flash drives and can use the shitty NAND elsewhere...
  • jabber - Tuesday, June 25, 2019 - link

    Kingston are my go to budget SSD brand. I bought dozens of those much moaned at V300 SSDs in the day. Did I care? No, because they were light years better than any 5400rpm pile of junk in a laptop or desktop.

    The other reason? Not one of them to date has failed. Including the V400 and onwards.

    They may not be the fastest (what's 30MBps between friends) but they are solid drives.

    Nothing more boring than a top end enthusiast SSD that is bust.

    Recommended.
  • GNUminex_l_cowsay - Tuesday, June 25, 2019 - link

    This whole article raises a question for me. Why is SATA still locked into 6Gbps? I get that there is an alternative higher performance interface but considering how frequently USB 3 has had its bandwidth upgraded lately it seems like a maximum bandwidth increase should be reasonable.
  • thomasg - Tuesday, June 25, 2019 - link

    There's just no point in updating SATA.
    6 Gbps is plenty for low-performance systems, SATA works well, is cheap and simple.

    For all that need more performance, the market has moved to PCIe and NVMe in their various form factors, which is just a lot more expensive (especially due to the numerous and frequently changed form factors).

    USB, as not only an, but THE external port that all users are facing has a lot more pressure behind it to get updated.
    Users touch USB all the time, there's demand for a lot of things over USB; most users never touch internal drives (in fact, most users actively buy hardware without replaceable internal drives), so there's no point in updating the standard.
    The manufactures can just spin new ports and new connectors, since they ship only complete systems anyway.
  • Dug - Tuesday, June 25, 2019 - link

    "There's just no point in updating SATA."
    That could be said for USB, pci, etc.
    There is a very good reason to go beyond an interface that is already saturated, and it doesn't have to be regulated to low performance systems.
  • Samus - Wednesday, June 26, 2019 - link

    SATA is an ancient way of transferring data. Why have a host controller on the PCI BUS when you can have a native PCIe device like NVMe. Further, SATA even with AHCI simply lacks optimization for flash storage. There doesn't seem to be an elegant way of adding NVMe features to SATA without either losing backwards compatibility with AHCI devices or adding unnecessary complexity.
  • TheUnhandledException - Saturday, June 29, 2019 - link

    SATA the protocol was built around supporting spinning discs. Making it work at all for solid state drices was a hack. A hack with a lot of unnecessary overhead. It was useful because it provided a way to put flash drives on existing systems. Future flash will will NVMe over PCIe directly. The only reason for upgrading SATA would be if hard drives actually needed >600 MB/s and they likely never will. So while we will have faster and faster interfaces for drives it won't be SATA. It would be like saying well because we made HDMI/DP faster and faster why not enhance VGA port to support 8K. I mean in theory we could but VGA to support a digital display is a hack and largely just existed for backwards and forwards compatibility because of analog displays.
  • MDD1963 - Tuesday, June 25, 2019 - link

    Why limit yourself to 550 MB/sec? I think having 6-8 ports of SATA4/SAS spec (12 Gbps) would breathe new life into local storage solutions...(certainly a NAS would be limited by even 10 Gbps networks, however, so equipped, but,..gotta start somewhere with incremental improvements, and, many SATA3 spec drives have now been limited to 500-550 MB/sec for years!)
  • Spunjji - Wednesday, June 26, 2019 - link

    You kinda covered the reason right there - where the performance is really needed, SAS (or PCIe) is where it's at. There really is no call for a higher-performing SATA standard.

Log in

Don't have an account? Sign up now