Mixed 4KB Random

In real world, workloads are rarely just pure reads or writes, hence it's important to test mixed performance as it better illustrates the performance under an enterprise workload. The read/write distribution varies greatly depending on the workload, but 70% read and 30% write is often considered as the benchmark for mixed performance. It's a little write heavy to mirror the most read-centric workloads (like media streaming or cloud storage), but it's fairly realistic for mimicking virtual desktop infrastucture (VDI) workloads for example.

The test sequence is similar to the random write benchmark. I start off with a two-hour sequential write pass, which is followed by six hours of 4KB random IOs at QD32 with 70% being reads and 30% writes. I again record the results of the last 500 seconds to ensure that the results represent steady-state performance. I also test queue depth scaling after the six-hour run and as a final test I run a 4KB random IO test (QD32) at six different read/write distributions in order to determine performance for different workloads. 

Mixed 4KB Random 70R/30W (QD32) Performance

The 70R/30W mixed performance is a slight disappointment. While the M510DC deliver considerably higher random write performance than the 845DC EVO, it cannot match the EVO in mixed performance. 

At low queue depths the M510DC actually provides better performance than the EVO, but after QD8 the performance no longer scales optimally, which is inherent in the design as the same scaling phenomenon is present in the M500DC as well. 

The performance at different read/write distributions is overall good and on par with the CloudSpeed Eco and 240GB EVO. For performance-focused mixed workloads, the M500DC is a better option, although ultimately the S3700 easily takes the crown here. 

Mixed 4KB Random 70R/30W (QD32) Consistency

Consistency again leaves something to be desired because compared to the EVOs the M510DC isn't a very consistent drive. 

Mixed 4KB Random 70R/30W (QD32) Efficiency

In power efficiency, however, the M510DC is very competitive and considerably better than the CloudSpeed Eco.

Micron M510DC
Default

The M510DC is consistently inconsistent. It's not as bad as the M500DC, but given that the 845DC EVO provides better performance at much higher consistency it's hard to recommend the M510DC for applications that require very consistent mixed IO performance.

Micron M510DC
Default

The same is also visible in the latency distribution. While the M510DC has frequent IOs in the range of 100µs, the consistency is again damaged by 3% of >10ms IOs. No other drive has IOs in that range, so it's a bit alarming given how even the CloudSpeed Eco consistently delivers IOs below 10ms. 

Micron M510DC
Default
4KB Random Write 4KB Random Read
Comments Locked

22 Comments

View All Comments

  • Kristian Vättö - Tuesday, July 21, 2015 - link

    The endurance rating is basically the warranty for the drive, so it's a quite practical measure. If the drive fails before exceeding the rating, then you are eligible for replacements.
  • nils_ - Friday, July 24, 2015 - link

    It's a bit pointless to have a warranty on DC hardware if you have to send in your broken drive to get a replacement, since you're supposed to destroy it.
  • Kristian Vättö - Sunday, July 26, 2015 - link

    I'm not sure how exactly enterprise warranties work, but I would expect them to be more flexible given that the customers often do business worth of at least hundreds of thousands of dollars. Besides, with encryption the data is safe even when sent to the manufacturer.
  • toyotabedzrock - Wednesday, July 22, 2015 - link

    The 845dc pro was not in the mixed workload.
  • Rekkx - Wednesday, July 22, 2015 - link

    The problem with the Samsung 845DC (EVO and Pro) is that it is already EOL.
  • andjohn2000 - Wednesday, July 22, 2015 - link

    Micron SSD is not reliable and can easily corrupt the data
  • ZeDestructor - Friday, July 24, 2015 - link

    Source? Cause I've seen nothing that indicates they're any better or worse than the comnpetition.
  • ZeDestructor - Friday, July 24, 2015 - link

    SandForce SF-2281 excepted....
  • velanapontinha - Friday, July 24, 2015 - link

    Hi, Kristian.

    Long time reader, here (since '98, I think), shy poster, though.

    I wonder if you guys would consider evaluating real-life endurance in enterprise SSDs. Much like this:

    http://blog.innovaengineering.co.uk/

    Cheers,

    Fernando
  • Kristian Vättö - Sunday, July 26, 2015 - link

    The problem with evaluating endurance, especially on enterprise drives, is that by the time we have any useful data to show the drives are already obsolete. A sample size of one isn't enough either for any statistical analysis, so to really test endurance in real world we would need our own mini data center with hundreds of drives to get sufficient amount of data.

Log in

Don't have an account? Sign up now