Skip to main content

Samsung 845DC EVO SSD Review: 3-Bit MLC Hits The Enterprise

Samsung Introduces 3-Bit MLC NAND To The Enterprise

Samsung took a fairly straightforward approach to the introduction of its 845DC EVO: start with the 840 EVO consumer SSD, add power-loss protection, and tweak the firmware for enterprise workloads. Many other storage vendors have gone down a similar path. So, this re-purposing isn't out of the ordinary. Even still, I want to commend Samsung's execution.

Achieving over 14,000 sustained steady-state write IOPS, the 845DC EVO performs consistently well. But what really caught my eye was read performance. We easily hit Samsung's 87,000 IOPS specification. Sequential read throughput is also excellent. Best-in-class, even. Given the price range the 845DC EVO is expected to occupy, you won't find a better performer. Performance is truly maximized for this real-focused target market.

As with every enterprise buying decision, use case and price are both going to be major factors in determining whether Samsung's 845DC EVO is the right SSD. The drive should show up for well under $1/GB, and the closer that Samsung gets to 840 EVO pricing, the better. For its intended purpose of read workloads, you just can't beat the speed. If your application relies on any amount of write performance, pricing will weigh more heavily on the decision. Should that be the case, there are other enterprise-class SSDs like Micron's M500DC hovering around $1.15/GB offering superior write performance.  

[UPDATE - Samsung reached out to us after publication confirming that the 845DC EVO 240, 480 and 960 GB models will be priced at $249.99, $489.99 and $969.20, respectively.  While this is higher than we expected, it is still one of the most cost effective enterprise SSDs on the market.  Expect that number to drift lower over time and with volume purchasing]

Even with, presumably, great pricing and performance, some customers will still shy away from the 845DC ECO simply because of its TLC, er, 3-bit MLC NAND. Samsung is betting that a year's worth of reviews and endurance testing on the 840 will alleviate those concerns. At this point, we believe that 3-bit MLC is ready for entry-level, read-focused enterprise use. The important point is to know your workload. If your application matches what the 845DC EVO can do, we think you'll be pleasantly surprised at how well the drive fares.

  • SteelCity1981
    So basicly it's the more durable version of the 840 evo much like opertons and xeons are to the FX and core i7 series.
    Reply
  • Plusthinking Iq
    like we know now after the ssd endurance test samsung is the worst enterprise candidate.
    Reply
  • drewriley
    13419610 said:
    So basicly it's the more durable version of the 840 evo much like opertons and xeons are to the FX and core i7 series.

    Yes, that's a fair analogy. Just like the Xeon E3-1275v3 is an i7-4770K, but with ECC support.

    Reply
  • damric
    I've yet to see an SSD fail due to read/write endurance. I only see them fail when the controller gets bugged, which seems to happen all the time, especially on loss of power.



    Reply
  • soundping
    I'm guessing this SSD doesn't have to new firmware code that extends life and speed.
    Reply
  • jase240
    Another win for the EVO. This SSD modified for enterprise workloads makes it a good buy for webservers.

    Hopefully the price will go down after launch, and then I see this being the best choice of webhosts.

    Cheaper and adequate for that workload.
    Reply
  • Nightmare Twily
    Eh I'll keep my 840 EVO 250GB
    Reply
  • Nuckles_56
    "Even still, I wand to commend Samsung's execution." (last page 1st paragraph) I guess that is supposed to be want, unless Drew Riley has become a wizard now :D
    Reply
  • Menigmand
    Commending their execution would be a bit harsh, don't you think?
    Reply
  • drewriley
    13426610 said:
    Commending their execution would be a bit harsh, don't you think?

    I'm sure worse things were said about Samsung at WWDC '14 yesterday ;)
    Reply