• "How Reliable are SSDs?"

    From Lynn McGuire@21:1/5 to All on Thu Feb 21 16:19:12 2019
    "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, are immune to magnetic fields, 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."

    "We’re now seeing SSDs with capacities that used to be reserved for HDDs
    and at prices that no longer make our eyes water. 500 GB SSDs are now affordable (under $100), and 1 TB drives are reasonably priced ($100 to
    $150). Even 2 TB SSDs fall into a budget range for putting together a
    good performance desktop system ($300 to $400)."

    "The bottom line question is: do SSD drives fail? Of course they do, as
    do all drives eventually. The important questions we really need to be
    asking are 1) do they fail faster than HDDs, and 2) how long can we
    reasonably expect them to last?"

    Lynn

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Lynn McGuire on Thu Feb 21 18:31:00 2019
    On 2/21/19 3:19 PM, Lynn McGuire wrote:
    1) do they fail faster than HDDs,

    It depends on the workload.

    SSDs are write sensitive. So anything that writes a lot of data will
    cause them to fail faster.

    Most SSDs are rated in how many times a day they can have the entire
    drive capacity written. Some drives are a fraction, some are single
    digit multiples. I don't think I've seen any that are two digit multiples.

    2) how long can we reasonably expect them to last?"

    Again, it depends on the workload.

    You should also be aware of the failure mode. Some drives fail such
    that they become read only. Others fail and become a brick. The former
    allows you to copy data off. The latter … well I hope you had good backups.



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Mark Perkins@21:1/5 to lynnmcguire5@gmail.com on Thu Feb 21 19:46:34 2019
    On Thu, 21 Feb 2019 16:19:12 -0600, Lynn McGuire
    <lynnmcguire5@gmail.com> wrote:

    "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, are immune to magnetic fields, 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."

    "We’re now seeing SSDs with capacities that used to be reserved for HDDs
    and at prices that no longer make our eyes water. 500 GB SSDs are now >affordable (under $100), and 1 TB drives are reasonably priced ($100 to >$150). Even 2 TB SSDs fall into a budget range for putting together a
    good performance desktop system ($300 to $400)."

    "The bottom line question is: do SSD drives fail? Of course they do, as
    do all drives eventually. The important questions we really need to be
    asking are 1) do they fail faster than HDDs, and 2) how long can we >reasonably expect them to last?"

    Good article, but they could have added one more paragraph to deal with
    the questions and differences around:
    2.5" SATA SSDs
    SATA m.2
    NVMe/PCIe m.2

    Maybe there will be a follow-up article at some point.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Mark Perkins on Thu Feb 21 21:08:48 2019
    On 2/21/19 8:59 PM, Mark Perkins wrote:
    I've never seen an SSD rated that way. I've seen TBW and "years" (the
    latter for warranty purposes), but never what you described. Is there
    a specific brand that describes 'life' that way? You said 'most', but
    I assume it's only one brand.

    Drive Writes Per Day.

    Link - Speeds, Feeds and Needs – Understanding SSD Endurance
    - https://blog.westerndigital.com/ssd-endurance-speeds-feeds-needs/

    § The SSD Endurance Equation



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Mark Perkins@21:1/5 to gtaylor@tnetconsulting.net on Thu Feb 21 21:59:54 2019
    On Thu, 21 Feb 2019 18:31:00 -0700, Grant Taylor
    <gtaylor@tnetconsulting.net> wrote:

    On 2/21/19 3:19 PM, Lynn McGuire wrote:
    1) do they fail faster than HDDs,

    It depends on the workload.

    SSDs are write sensitive. So anything that writes a lot of data will
    cause them to fail faster.

    Most SSDs are rated in how many times a day they can have the entire
    drive capacity written. Some drives are a fraction, some are single
    digit multiples. I don't think I've seen any that are two digit multiples.

    I've never seen an SSD rated that way. I've seen TBW and "years" (the
    latter for warranty purposes), but never what you described. Is there a specific brand that describes 'life' that way? You said 'most', but I
    assume it's only one brand.

    2) how long can we reasonably expect them to last?"

    Again, it depends on the workload.

    Well, everyone agrees that writing is more destructive than reading, so
    that's why SSD manufacturers provide a TBW rating, total terabytes
    written. The actual warranty can be written in such a way as to say that
    the drive is warranted until you reach the TBW or the years, whichever
    comes first.

    You should also be aware of the failure mode. Some drives fail such
    that they become read only. Others fail and become a brick. The former >allows you to copy data off. The latter … well I hope you had good backups.

    Backups are always a good idea. That didn't go away with SSDs.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Mark Perkins@21:1/5 to gtaylor@tnetconsulting.net on Thu Feb 21 23:50:21 2019
    On Thu, 21 Feb 2019 21:08:48 -0700, Grant Taylor
    <gtaylor@tnetconsulting.net> wrote:

    On 2/21/19 8:59 PM, Mark Perkins wrote:
    I've never seen an SSD rated that way. I've seen TBW and "years" (the
    latter for warranty purposes), but never what you described. Is there
    a specific brand that describes 'life' that way? You said 'most', but
    I assume it's only one brand.

    Drive Writes Per Day.

    Link - Speeds, Feeds and Needs – Understanding SSD Endurance
    - https://blog.westerndigital.com/ssd-endurance-speeds-feeds-needs/

    § The SSD Endurance Equation

    Ugh! So it's a [WD] blog where the guy cautions *against* using 'drive
    writes per day' because it's *not* a good metric for measuring drive
    endurance, due to the fact that it doesn't give apples to apples
    comparisons across drive capacities. He then goes on to say that TBW is
    a better metric, with which I (and virtually everyone else) agree.

    The question remains: is there an SSD manufacturer that uses 'drive
    writes per day' in their marketing materials to help customers figure
    out the endurance properties of their product? So far, no.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Mark Perkins on Fri Feb 22 12:42:01 2019
    On 2/21/19 10:50 PM, Mark Perkins wrote:
    Ugh! So it's a [WD] blog where the guy cautions *against* using 'drive
    writes per day' because it's *not* a good metric for measuring drive endurance, due to the fact that it doesn't give apples to apples
    comparisons across drive capacities. He then goes on to say that TBW is
    a better metric, with which I (and virtually everyone else) agree.

    That's just the first thing I came across in a quick search while tired
    brain dead.

    The question remains: is there an SSD manufacturer that uses 'drive
    writes per day' in their marketing materials to help customers figure
    out the endurance properties of their product? So far, no.

    I don't know about drive manufactures. But I do know that writes per
    day is the unit of measure that all of my colleagues and all of the
    vendors that we've been talking to for enterprise drives over the last
    18 months. Vendors such as:

    · Cisco
    · HP
    · Dell
    · Supermicro
    · Other white box vendors

    As I write that list I wonder if maybe it's server manufacturers /
    vendors / OEMs that use drive writes per day and not actual drive manufacturers.

    I'm not saying that drive writes per day is proper, just that I've seen
    it used a LOT more than total bytes written.



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From pbx5619@gmail.com@21:1/5 to All on Fri Mar 29 07:23:52 2019
    My 1st Transcend 370s 64gb lasted fr 11-15 to 3-17, 2nd unit ( warranty claim ) lasted fr 4-17 to 3-19, even with a big heatsink below its aluminium base plate.
    Every installation of Windows ( on different parts of storage area ) lasted fr 3 days to 4 mth maximum.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From pbx5619@gmail.com@21:1/5 to All on Sat Mar 30 00:15:18 2019
    My Apacer AS340 120 gb is lousy too : bought in 3-17, longest usable time for Win 8 installed ( @ front of storage*area ) was just 6 mth. I had to install Win 8 on different parts of * to try find a more durable location : some parts lasted just 5 days,
    now I use a location 82 gb from front ( = left side of Win 8 Disc Management chart ), I had to repair this partition 20x since 15-6-18 but it's still usable.
    6 data files stored turned bad, & I had to re-enter data, wasted ~4 hr.
    I put silver paste on its base plate & a ( AMD, Barton ) hsink ( with a fan ) below plate : hsink will warm up a bit if no fan cools hsink, this fact proves that heat can be extracted from a plastic plate too.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From pbx5619@gmail.com@21:1/5 to All on Sun Mar 8 01:01:33 2020
    Since Oct'19 I notice my Apacer SSD has fewer troubles ( unbootable unless repaired by either DVD or another SSD's Windows ), after I added cooling to mboard components : (i) fit a 3mm thick hsink onto cpu socket's resistors (ii) fit a 7mm long hsink to
    bottom of north bridge (iii) fit a AM2 hsink ( 2ndary ) onto north bridge's primary hsink ( silicone @ contact area ) (iv) reduce voltage to (AM3) cpu by 0.175 v ( cpu was too hot for this voltage without this 3mm hsink fitted ), north bridge by 0.175 v
    too, Adata DDR3 ram ( with hsink added ) to 1.36 v ( @ CL11, no tighter @ this voltage ). I presume south bridge too became cooler ( ; gets less heat fr cpu & n-bridge & ram ).
    Now the same space holding Win 8 ( used to last 6 mth max, since 4-18 ) has lasted 8 mth already !

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Yousuf Khan@21:1/5 to Lynn McGuire on Sat Apr 4 22:06:43 2020
    On 21/02/2019 5:19 PM, Lynn McGuire wrote:
    "How Reliable are SSDs?"
        https://www.backblaze.com/blog/how-reliable-are-ssds/

    I used to also think SSD's are more reliable than HDD's, but recently
    I've had nothing but trouble with a particular brand of SSD, the Adata
    SU630 series is absolutely crap. I've already had to return 3 of them,
    and I'm getting ready to return my 4th. Thank god I got good backups!
    They are good about exchanging their products, but I doubt that they've
    even noticed that I've returned 4 of their products already under
    warranty. They don't even ask questions, just take your RMA order. They
    must be using the worst Flashram in the world, from the reject pile of
    every manufacturer around. I've had enough, and I'm going to replace
    with a WD SSD now, but I will get the latest replacement and probably
    put it into an external case for occasional large storage requirements
    that don't need to be on all of the time.

    Yousuf Khan

    --
    Sent from Giganews on Thunderbird on my Toshiba laptop

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From VanguardLH@21:1/5 to Yousuf Khan on Sun Apr 5 00:50:55 2020
    Yousuf Khan <bbbl67@spammenot.yahoo.com> wrote:

    Lynn McGuire wrote:

    "How Reliable are SSDs?"
    https://www.backblaze.com/blog/how-reliable-are-ssds/

    I used to also think SSD's are more reliable than HDD's, but recently
    I've had nothing but trouble with a particular brand of SSD, the Adata
    SU630 series is absolutely crap. I've already had to return 3 of them,
    and I'm getting ready to return my 4th. Thank god I got good backups!
    They are good about exchanging their products, but I doubt that they've
    even noticed that I've returned 4 of their products already under
    warranty. They don't even ask questions, just take your RMA order. They
    must be using the worst Flashram in the world, from the reject pile of
    every manufacturer around. I've had enough, and I'm going to replace
    with a WD SSD now, but I will get the latest replacement and probably
    put it into an external case for occasional large storage requirements
    that don't need to be on all of the time.

    I was expecting the article to provide some actual statistics,
    especially since the author was Backblaze. Instead it was just a bunch
    of general information with no statistics at all. Pretty useless since
    it never does address how reliable are SSDs as experienced from actual
    use in their data centers.

    However, Backblaze doesn't use SSDs for storage of customer data, just
    for a few boot drives or as frontend servers, like database servers.
    They don't have many to provide any statistics, so they won't have any statistics to report. Yet that article is just generalized fluff about
    SSDs versus HDDs. You cannot draw many conclusions from it, and nothing substantial regarding reliability.

    If you want to increase the lifespan (aka endurance) of an SSD, increase
    its overprovisioning. That allocates more reserved space to accomodate
    failed memory blocks that will happen eventually. You lose some
    capacity for the unallocated space on the SSD for more (well, any) overprovisioning, but if you're getting tight on space (and aren't
    collecting tons of garbage files or data that could be stored elsewhere
    like on a cheaper HDD) then you really should get higher or more drives.

    Most consumers look at the marketing data, like capacity. Important is
    the read and write speed (with writes being slower than reads because of
    the procedure to do writes). I see the same for most buyers of USB
    thumb flash drives: they go for capacity without ever investigating how
    fast (or how slow) they are, but then many makers don't publish those
    specs.

    https://www.youtube.com/watch?v=Q15wN8JC2L4 https://www.seagate.com/tech-insights/ssd-over-provisioning-benefits-master-ti/ https://www.kingston.com/us/ssd/overprovisioning

    I think the typical overprovisioning is 10% of the rated capacity of the
    SSD, but I think the amount varies by capacity with 10% used for all consumer-grade drives over some threshold in capacity. The server-grade
    SSDs usually have 20% overprovisioning, and that's what I use, too,
    although my SSDs don't get anywhere the volume of writes that business
    use would encounter.

    I've stuck with Samsung for SSDs: both as encased drives for internal
    use connected to power and SATA cables from the motherboard (Samsung 850
    EVO 2.5" 250 GB SATA-3, bought April 2016) used in my prior Win7 box,
    and for m.2 drives into sockets on the mobo (Samsung 970 Pro M.2/2280 1
    TB Gen3 NVMe PCI 2-bit MLC, bought April 2019) used in my latest build
    (Win10). Never had any problems with those. I might use Crucial (who
    doesn't make anything, but has good specs with the actual plants) or
    Crucial (their high end products).

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Yousuf Khan@21:1/5 to VanguardLH on Sun Apr 5 11:22:09 2020
    On 05/04/2020 2:50 AM, VanguardLH wrote:
    I was expecting the article to provide some actual statistics,
    especially since the author was Backblaze. Instead it was just a bunch
    of general information with no statistics at all. Pretty useless since
    it never does address how reliable are SSDs as experienced from actual
    use in their data centers.

    However, Backblaze doesn't use SSDs for storage of customer data, just
    for a few boot drives or as frontend servers, like database servers.
    They don't have many to provide any statistics, so they won't have any statistics to report. Yet that article is just generalized fluff about
    SSDs versus HDDs. You cannot draw many conclusions from it, and nothing substantial regarding reliability.

    Yeah, I agree, it's just an introductory piece on what an SSD is, and
    that's all. Plenty of those articles already.

    If you want to increase the lifespan (aka endurance) of an SSD, increase
    its overprovisioning. That allocates more reserved space to accomodate failed memory blocks that will happen eventually. You lose some
    capacity for the unallocated space on the SSD for more (well, any) overprovisioning, but if you're getting tight on space (and aren't
    collecting tons of garbage files or data that could be stored elsewhere
    like on a cheaper HDD) then you really should get higher or more drives.

    What a really useful metric for SSD's would be is what is their proper operating temperatures? They are much higher than HDD's, but how much
    over are they? Each manufacturer seems to have its own ideas, and then
    many of them don't even release that info. The aforementioned crap that
    I've been having so much trouble with, the Adata SU600-series, seem to
    overheat at in their 50's and 60's, and I'm finding that it's regularly operating at the mid-50's! I don't officially know that their limit is
    the 50's/60's, that's just what I've discovered over a couple of years
    of returning them over and over again. Adata don't release their own
    specs about this, probably because they know that most of their drives regularly operate above this level, which would result in even more
    returns when worried owners see it running that high.

    Yousuf Khan

    --
    Sent from Giganews on Thunderbird on my Toshiba laptop

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