• =?UTF-8?B?4oCcU1NEIDEwMTogSG93IFJlbGlhYmxlIGFyZSBTU0RzP+KAnQ==?=

    From Lynn McGuire@21:1/5 to All on Thu Mar 30 16:21:39 2023
    XPost: alt.comp.hardware.pc-homebuilt

    “SSD 101: How Reliable are SSDs?”
    https://www.backblaze.com/blog/how-reliable-are-ssds/

    “What’s not to love about solid state drives (SSDs)? They are faster
    than conventional hard disk drives (HDDs), more compact, have no moving
    parts, use less electricity, and can withstand more shocks and vibration
    than conventional magnetic platter disks. And, they are becoming
    available in larger and larger capacities while their cost comes down.”

    "In this post we’re going to consider the issue of SSD reliability.
    We’ll take a closer look at:
    1. SSD tech.
    2. SSD storage memory.
    3. Reliability factors.
    4. Signs of SSD failure."

    Lynn

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Paul@21:1/5 to Lynn McGuire on Thu Mar 30 21:13:24 2023
    XPost: alt.comp.hardware.pc-homebuilt

    On 3/30/2023 5:21 PM, Lynn McGuire wrote:
    “SSD 101: How Reliable are SSDs?”
       https://www.backblaze.com/blog/how-reliable-are-ssds/

    “What’s not to love about solid state drives (SSDs)? They are faster than conventional hard disk drives (HDDs), more compact, have no moving parts, use less electricity, and can withstand more shocks and vibration than conventional magnetic platter
    disks. And, they are becoming available in larger and larger capacities while their cost comes down.”

    "In this post we’re going to consider the issue of SSD reliability. We’ll take a closer look at:
    1. SSD tech.
    2. SSD storage memory.
    3. Reliability factors.
    4. Signs of SSD failure."

    Lynn

    These are apples and oranges for Backblaze, since
    SSDs are not cheap, and they have different characteristics
    on write.

    A 30TB SSD costs $3800 or so. A 22TB HDD might cost $700
    and you'd get more usage from it while doing backups.

    SSDs could last forever, if they could find a way to
    anneal them while in service. Which isn't likely to happen.
    The effect has been observed in the lab, but so far, no
    one has created a device with that capability.

    The 3D TLC ones on the end of the graph, have a higher replacement rate.

    https://www.usenix.org/conference/fast20/presentation/maneas

    And I see the dude is learning, you should not put identical
    SSD make and model drives into RAID1 mirrors. Quelle suprise.
    The failures correlate. So what you should be doing, is mixing
    your drives, like a Sabrent 4TB and a Samsung 4TB, and then
    the second failure (RAID degrade to RAID fail transition) won't
    correlate quite so tightly. Or you could even mix sizes, so that
    the TBW difference between drives, helps decorrelate failure a tiny bit.

    Firmware does make a difference, as we were observing that with
    first generation SSDs (terrible firmware), and just recently,
    some sort of issue popped up with 980 and 990 drives (you have to
    flash them to correct this, and do it "before the drive burns out").
    The automatic sparing which is going on in error, is not reversible
    on those drives, so the faster you flash a newer firmware, the
    better it is for the drive.

    If your computer has a UPS, and the automatic shutdown cable is
    connected, I would expect this to measurably improve SSD "qualities".

    Paul

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)