Worst-Case Performance Consistency

The performance of the storage bridges / drives in various real-world access traces as well as synthetic workloads was brought out in the preceding sections. We also looked at the performance consistency for these cases. Power users may also be interested in performance consistency under worst-case conditions.

Flash-based storage devices tend to slow down in unpredictable ways when subject to a large number of small-sized random writes. Many benchmarks use that scheme to pre-condition devices prior to the actual testing in order to get a worst-case representative number. Fortunately, such workloads are uncommon for direct-attached storage devices, where workloads are largely sequential in nature. Use of SLC caching as well as firmware caps to prevent overheating may cause drop in write speeds when a flash-based DAS device is subject to sustained sequential writes.

Our Sequential Writes Performance Consistency Test configures the device as a raw physical disk (after deleting configured volumes). A fio workload is set up to write sequential data to the raw drive with a block size of 128K and iodepth of 32 to cover 90% of the drive capacity. The internal temperature is recorded at either end of the workload, while the instantaneous write data rate and cumulative total write data amount are recorded at 1-second intervals.

Sequential Writes to 90% Capacity - Performance Consistency
TOP: BOTTOM:

The SanDisk Extreme PRO v2 4TB maintains peak performance of around 1850 MBps for around 30 seconds before moving down a bit to a respectable 1600 MBps for the rest of the workload. The numbers point to a SLC cache of around 60GB, but the direct TLC write has very goood performance too. The temperature delta between the start and the end was 15C, but the maximum temperature was only 56C pointing to a very effective thermal solution.

The Crucial X6 maintains peak advertised performance of around 780 MBps for more than 16 minutes, pointing to a SLC cache of around 787 GB. However, once the cache runs out, the speeds move down to around 75 MBps till around 1.47TB of total continuous data writes. Beyond that, the speeds go down to around 50 MBps for the rest of the workload. The temperature delta is around 61C, but the slow writes in the latter part of the process means that not much heat gets generated anyway. The thermal solution with the plastic casing for the SSD is acceptable.

PCMark 10 Storage Bench - Real-World Access Traces Power Consumption and Concluding Remarks
POST A COMMENT

20 Comments

View All Comments

  • arashi - Friday, August 20, 2021 - link

    That's for the 4TB one, but for the 1TB variant it's still 200GB, which is plenty big. Reply
  • repoman27 - Thursday, August 19, 2021 - link

    @ganeshts why aren’t the results from your recent review of the Silverstone MS12 and Yottamaster HC2-C3 USB 3.2 Gen 2x2 20Gbps enclosures included in these charts?

    I’d be very interested to see how a roll your own solution with SK hynix Gold P31 fares against the SanDisk Extreme V2 and Extreme PRO V2. This would be especially useful at the 2TB capacity level, as WD seems to be charging a $100 premium for the 4TB models, and the Gold P31 maxes out at 2TB now.
    Reply
  • ganeshts - Thursday, August 19, 2021 - link

    Yes, valid question. I thought about it a bit while preparing the graphs, and finally felt that the gulf in the capacity points (1TB vs. 4TB) could lead to readers misunderstanding the relative merits of the two platforms. So, decided to focus on the 'high-capacity' aspect alone.

    I do have the review of the Seagate FireCuda Gaming SSD (20Gbps) 1TB coming up soon, will be including the storage bridge results in that for comparison.

    As you mention, 2TB is fast becoming the mainstream offering in terms of capacity as 3D NAND layer counts keep going up. I will try to add the new testbed results from the old 2TB samples in future reviews.
    Reply
  • flgt - Thursday, August 19, 2021 - link

    I’ve been really happy with my 2TB SanDisk. It breathed new life into my Xbox One X when I realized I’d be stuck with it for the foreseeable future. Reply
  • COtech - Thursday, August 19, 2021 - link

    How often do these portable SSDs need to be powered to reliably retain data ?

    Are they a good archive solution?
    Reply
  • Tams80 - Thursday, August 19, 2021 - link

    Unless you're moving them at lot (or will take them on a rough journey), then I think it's a toss-up.

    As always, redundancy is important and for that Blu-ray archival disks are a great choice as one of the back-ups.
    Reply
  • data21 - Tuesday, September 14, 2021 - link

    Nicely written blog on SanDisk Extreme PRO and Crucial X6 4TB Portable SSDs comparisions. for data recovery related issued in SSD can reach out to lifeguard data recovery services Reply
  • jabakobob - Wednesday, November 3, 2021 - link

    I'd like to correct a major error in this review: The X6 does not have 800GB of SLC cache.

    Instead, it just uses empty QLC cells as cache and writes only 1 bit to them instead of 4. So it effectively has a big SLC cache as long as it is <20% full. But the fuller it gets, the less cache it has!

    So basically it's a nice and fast drive as long as you only use 800GB. If you use more than that, performance craters.

    If you buy this drive to store a lot of data, it will never perform as well as in the tests. I assume most of the tests (except the worst case test) were done with an empty drive, so they are pretty much useless. It would be interesting how the drive would perform if it is 50% full or 80% full.

    If you actually need 4TB of capacity, don't buy this drive.

    Source: https://www.computerbase.de/2021-03/crucial-x6-por...
    Reply
  • chemist1 - Friday, August 26, 2022 - link

    In order to make these tests as real-world as possible, you should first add files to the disks until they are about 50%-60% full (as opposed to running them on empty disks). Do you do that? Reply
  • yifu - Monday, October 10, 2022 - link

    I bought the x6 2tb last month, a return item for 110€. Some 500gb write on disk.
    I loved it. It Use very little power. I don’t think you test on the new firmware. Plugged in iPad Pro all day forgot to unplug. Didn’t even use 1% of batterie. I guess host device sleep, x6 drove no power. Great product for its targeted users environment.
    Reply

Log in

Don't have an account? Sign up now