4 ms·
The QN322 is a garbage-tier drive, equivalent to the QLC-refreshed Crucial P2 that "averaged USB 2.0-like speeds of just 40 MBps after the SLC cache was full."[
by tbyehl 4y ago
The QN322 is a garbage-tier drive, equivalent to the QLC-refreshed Crucial P2 that "averaged USB 2.0-like speeds of just 40 MBps after the SLC cache was full."[1] Lots of value drives out there for $10-$30 more than won't perform worse than a mechanical hard drive after exhausting SLC.
[1] https://www.tomshardware.com/features/crucial-p2-ssd-qlc-flash-swap-downgrade/2 https://www.tomshardware.com/features/crucial-p2-ssd-qlc-fla...
- wtallis 4y agoThat "40 MBps after the SLC cache was full" was measured on a 500GB drive. A 2TB drive would have an SLC cache at least 4x larger and post-cache write speeds around 4x higher—bringing it up to roughly the level of mechanical hard drives and beyond the speed of Gigabit Ethernet. In short, the problem you're complaining about wouldn't matter for the drive in question.
- ksec 4y agoThat seems like a perfect SSD for NAS. Do we still need things like Data Scrubbing on SSD?
- wtallis 4y agoYes, data scrubbing (read only) is still a good idea for SSDs. If data in a marginal memory cell is starting to degrade, reading that data will cause the SSD to notice a correctable error and write a fresh copy of the data. Simply leaving the drive on the shelf would put you at more risk of degradation leading to an error that the drive was not able to correct (though a RAID layer above the individual drives may still be able to recover the data). Many drives will do some amount of background reading on their own to scan for degraded data. But this isn't visible to the host system and not guaranteed, so it shouldn't be relied upon and doing some extra reads (eg. once a month) is harmless. I strongly believe QLC SSDs are a great choice for NAS usage, unless it's very frequently modified data. It's still more expensive than hard drives, but the noise, power and performance benefits can be worth it.
- ksec 4y agoThanks. At $69 / 2TB it really is cheap. The only problem is NAS dont support M.2.
- tbyehl 4y agoThe Tom's review of the revised Crucial P2 500GB was a 2-chip and the QN322 2TB also appears to be 2-chip vs 1-chip for smaller capacities of the QN322, so the performance boost from higher capacity won't be there. And the SLC cache size isn't fixed but rather a fraction of the free space. There are also customer reviews of the 2TB on the Microcenter site noting that post-cache writes fall to 40MB/s. Back when QLC first took over the "value-tier" of the major brands I bought dozens of Inland drives because they were still TLC, making them better performing with higher endurance than other cheap SSDs, but the QN322 is a worst-of-QLC drive. Depending on the week it's only marginally cheaper than better drives — Intel 670p 2TB was $79.99 in-store two weeks ago.
- wtallis 4y ago> The Tom's review of the revised Crucial P2 500GB was a 2-chip and the QN322 2TB also appears to be 2-chip vs 1-chip for smaller capacities of the QN322, so the performance boost from higher capacity won't be there. You need to make the distinction between how many packages of NAND flash are soldered to the PCB, and how many dies of NAND are stacked in one BGA package. A 2TB SSD requires more than two NAND dies because the largest NAND dies in production are in the ballpark of 1-1.33Tbit, so a 2TByte drive would typically have 16 NAND dies in total. > And the SLC cache size isn't fixed but rather a fraction of the free space. That doesn't stop a 2TB drive from generally having 4x more SLC cache than a 500GB drive, other things being equal.
- tbyehl 4y agoThe published specs don't tell us how many layers of NAND there are, but they do claim the 2TB is only 1.8X the write performance of the 500GB and 1.125X the 1TB. Not the 4X you'd speculated. If you have a QN322, feel free to provide benchmark results.
- wtallis 4y agoNAND layers is another question entirely: that's about how many layers of memory cells each die has, not about how the dies are packaged together. And of course the performance specs for the 2TB don't show it being 4x faster than the 500GB model: those specs are for peak performance when using the SLC cache. Consumer SSD specs almost never list post-cache performance. I don't have one of the drive in question, but unlike you I actually know enough about how these things work to know what performance claims are plausible or not. (See eg. https://www.anandtech.com/show/16681 https://www.anandtech.com/show/16681 https://www.anandtech.com/show/16577 https://www.anandtech.com/show/16577 https://www.anandtech.com/show/16491 https://www.anandtech.com/show/16491 )