• How to Keep my SSD Alive for Longer

    Home » Forums » AskWoody support » PC hardware » PC hardware-General Questions » How to Keep my SSD Alive for Longer

    Tags:

    Author
    Topic
    #1922181

    I’m in the middle of a Windows system upgrade, but I’ve hit a wall deciding how to configure my storage. The basic problem is to do with the SSD:  are the steps I’m taking to extend its life even necessary?

    I could really use some advice from some hardware experts that understand how SSDs work in the real world.

    What I know is that SSDs are write-limited.  So the idea is to shift write-intensive activities off the SSD and onto normal hard drives.

    Obviously, putting a swap file on an SSD is a terrible idea, but what about all those other files that Windows is constantly creating and revising:  the Registry, the TEMP directory, Application Data, etc?

    On my current, Windows 7 system, I use a hard link to redirect my Users directory off the SSD (C:) and onto a 7200rpm hard drive (D:).  I then also keep all my documents on D:  So my system looks like this:

    • 120GB SSD – C: Windows 7 system files and applications (no swap file)
    • 750GB 7200rpm HDD – D: Users directory and other documents and temp files
    • also on the 750GB HDD – E: Software archive
    • 4TB HDD – F: Archive of Music, video recordings, DVD backups, etc (mostly large files)
    • 3TB 7200rpm HDD – G: Games, and overflow from the video archive

    The new system will have a 1TB SSD for Windows, applications and games, including Steam.

    Ultimately, I need to decide whether to leave the Users directory on the SSD as well, or move it to a regular hard drive (probably a partition on what is currently G:, since that 750GB drive is over 11 years old).

    Of course, the second issue is, how do I do that?  Using a hard link won’t work in Windows 10.  I gather you have to modify the OS install script to put \Users and \Program Data on a different drive.  I could really use a pointer to some instructions how to do that!

    Thanks in advance for any expert opinion on both of these questions!

    Viewing 8 reply threads
    Author
    Replies
    • #1922231

      What I know is that SSDs are write-limited

      SSD with normal use will out live several PCs replacements (by that time your will replace HDDs x100 times) so you have nothing to worry about nor take any special steps to “prolong its life”.
      You can write swap, hib, sleep files on the SSD without a worry.

      The life span of a Samsung 850 PRO with 1TB then results in:

      https://www.compuram.de/blog/en/the-life-span-of-a-ssd-how-long-does-it-last-and-what-can-be-done-to-take-care/

      • This reply was modified 5 years, 7 months ago by Alex5723.
      1 user thanked author for this post.
    • #1922483

      SSD’s may outlive an HDD but they will slow down eventually if they are not cleaned up with a “Trim” program.  I think Windows 10 has Trim, maybe you could find one for Win 7.

      I put a Samsung 250 GB 860 EVO SSD in my old Sony laptop that I have Linux Mint installed on, and linux has FSTrim that I use about every two weeks or so.

      Being 20 something in the 70's was far more fun than being 70 something in the insane 20's
    • #1923077

      I bought a 128GB Samsung 840 Pro SSD for my desktop PC 5.5 years ago, just upgraded to Windows 7 at the time.  I’ve never taken it easy on the drive… I’ve had my swap/page file on there since the beginning, and in those days I used to hit that page file very hard.  It was during the worst part of Firefox’s memory leaking days, and I tend to have tons of tabs open.  The speed of the SSD masked the immediate slowdown that let me know it was time to restart Firefox, so it would just continue to pound away at the SSD.  When I later checked it with Sysinternals’ RamMap, I saw that the system was having to repurpose (page out) into the top few priority categories, which is generally a sign that the RAM installed is way too small for the task at hand.

      I did buy more RAM for that PC, but the nature of memory leaks is that they just keep going, using up more and more memory, until something happens to stop it.  It delayed the memory crisis, but the page file still got hit hard from time to time.

      That SSD was the home to Windows (7 first, then 10 to be able to give it a fair tryout on my own well-known hardware, then back to 7, then 8.1, and now to Linux), and I’ve moved all kinds of stuff on and off that drive.  I’ve restored from backups more times than I can remember (all part of a normal computing day for me).

      I think it is fair to say that I have tortured that little SSD more than the vast majority of home users would.  And now, after 5.5 years, the drive reports that it still has 68 percent of its rated service life remaining.  But even when it reaches zero rated life remaining, it’s not the end… TechRepublic did a series of tests on SSDs, doing continuous writes on them until they died, and the 840 Pro was only halfway to its own death when it ran out of rated life.  At the very least, I have 11 more years of life left… but in reality it will be more, since Firefox (upon which Waterfox is based) has improved greatly, and most days Linux doesn’t even touch the swap file.

      The 840 I have is a pro series SSD, and the pro series from Samsung have longer lives than the Evos of the same generation (840 in this case), but each new generation increases the life of Evo and Pro alike.  I think I read somewhere that the NAND cells in my 850 Evo (now in my Dell G3 laptop) have a similar write-cycle rating as those in my my 840 Pro, and the 860 Evo in my Acer Swift is even better than the 850.

      On top of that, my 850 and 860 are both 1TB models, which increases their nominal service life 8 times compared to the 128GB 840 Pro.  There are two ways to improve the nominal service life of a SSD… increase the number of write cycles each NAND cell can tolerate before dying, or increase the number of NAND cells and spread the workload out so each cell gets fewer writes.

      The price I paid for my 128GB 840 Pro back then would now get you a 860 Evo 1TB, and it would have a significantly longer life than my 840 Pro… which is only a third done into its 16.5 year long rated life!

      I would definitely suggest that you do put the swap file on the SSD, as the SSD is far faster, and you want the most speed you can get for that.

      As for TRIM– my Windows 7 did this by itself after each write.  I did the TRIM test (a small command line utility) where the program writes a bunch of data to a given logical sector of the drive, then deletes it, and when you run it again, it attempts to read the data.  If the deleted data is still there, TRIM did not take place, as the purpose of TRIM is to finalize deletions and free up the NAND blocks.  If the data is gone, TRIM took place.

      In each case that I tried it, the data was gone on my Win 7 system.  I only ran the test, waited a minute, then ran it again, but it had trimmed during that time.

       

      Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
      XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
      Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

    • #1923159

      Ascaris: ” And now, after 5.5 years, the drive reports that it still has 68 percent of its rated service life remaining.

      This sounds like something useful to know. How does one interrogate an SSD to get that information?

      Ex-Windows user (Win. 98, XP, 7); since mid-2017 using also macOS. Presently on Monterey 12.15 & sometimes running also Linux (Mint).

      MacBook Pro circa mid-2015, 15" display, with 16GB 1600 GHz DDR3 RAM, 1 TB SSD, a Haswell architecture Intel CPU with 4 Cores and 8 Threads model i7-4870HQ @ 2.50GHz.
      Intel Iris Pro GPU with Built-in Bus, VRAM 1.5 GB, Display 2880 x 1800 Retina, 24-Bit color.
      macOS Monterey; browsers: Waterfox "Current", Vivaldi and (now and then) Chrome; security apps. Intego AV

      • #1923389

        The easiest way (generally only possible in Windows) is to use the utility that came with (metaphorically speaking, since hardware hardly ever comes with actual media anymore) the SSD.  In Samsung-land, it’s called Magician.

        From Linux, or from Windows if you don’t want to install Magician, you can get it from the SMART data.  On my Samsung SSDs as well as one SK-Hynix SSD I have, attribute ID 177, wear-leveling count, shows the percentage of rated life left in the drive (in the “normalized” column if you use Mint’s “Disks,” aka GNOME Disks).

        My Sandisk SSD apparently uses ID 232, endurance remaining, though for some reason it is not being populated with any data.

        On other units, it should be in SMART somewhere.

        Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
        XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
        Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

      • #1924018

        There’s just one little thing…

        The SMART or whatever lifecycle counter isn’t very reliable, particularly on older models. When I see broken SSDs, they’ve often “jumped” ahead, or failed with symptoms consistent with excessive wear with something like 30% lifetime remaining. (Could well be that the actual component that breaks isn’t the storage cells, but something else…)

        Also production runs being once-and-done doesn’t help much – by the time proper field reliability figures are even theoretically available, the same disk models aren’t available new any more.

        • #1924889

          When I see broken SSDs, they’ve often “jumped” ahead, or failed with symptoms consistent with excessive wear with something like 30% lifetime remaining.

          It could be heat (often seen in laptops) shortening the lives of the NAND cells, perhaps… or it

          (Could well be that the actual component that breaks isn’t the storage cells, but something else…)

          (emphasis added)

          That could be the case.  Other than NAND cells wearing out, the potential causes of SSD failure (often referred to as “logic board” failure) also apply to conventional hard drives.  The last hard drive failure I had, which I mentioned above, was a case of sudden death, where the SMART values looked great the day of the failure, then it just… died.  It was not possible to read any data from the drive at that point, but I do not remember if it was being recognized by the system.  I’d guess it was an electronics failure rather than anything mechanical in nature. That type of failure could have happened in a SSD too, even one whose NAND was still in good shape.

          There are all kinds of things that can happen to a drive out in the field that would shorten its life.  Excessive heat, poor quality power (ripple or voltage regulation), ESD damage during handling, moisture intrusion/corrosion, to list the ones that spring immediately to mind.

          In the TechReport series, there were six drives tested in the main series, plus a second example of one of the drives (The Kingston) to test its compression feature, which I don’t cite here.

          The Samsung 840 reached its rated end of life at 300 TB, and right at that point it had its first uncorrectible read error (which TR suggests as a sign that the drive needs to be replaced).  The drive continued to work without further uncorrectible errors until near its total failure at around 900TB.  Large-scale sector relocation started at the 100TB mark and increased linearly right to the end.

          The Kingston Hyper-X 3k (with the same uncompressible data as the other drives) reached its rated end of life at 728TB and died shortly thereafter.  Large-scale sector relocation had begun at 600TB and increased sharply, in a linear fashion, until the end.

          The Intel 335 reached its rated end of life at 700TB and bricked itself on that basis.  It only had a small number of relocated sectors, and showed no sign of imminent wear-out.  It was the only drive in the series designed to do this.

          The Corsair Neutron started reallocating sectors at 1100 TB, then died before it reached 1200.  The wear indicator indicated it had 75% life remaining at that point.  It was the only drive in the test that did not reach its rated end of life.

          The Samsung 840 Pro’s rated life ended at 700 TB, but the drive went on to live all the way until 2.4 petabytes.  The relocated sectors started to increase notably at 600TB and increased steadily (but not all that linearly) until the end.

          Only one of the drives, the Corsair, did not reach its rated end of life.  All of them except the Intel showed steadily increasing bad sectors before they failed completely, and that was because the Intel was the only one that committed suicide because of its end of rated life rather than dying naturally as the NAND cells actually wore out.  The Intel drive “died” so early that its NAND cells never had a chance to wear out.

          This is only six drives, of course, but the general trend is that they outlast their media wear indicator handily, and that they all show a steady and constant increase in bad sectors before they fail completely (with the Intel exception as noted above).  Some of them live a long time with steadily increasing bad sectors… the 840 Pro started having a lot of bad sectors at 600TB, but its life was only 25% over by then.  It lived three quarters of its life with steadily increasing bad sectors.

          This was in a controlled setting where the drives never got too hot, and where the age of the NAND cells was never able to become a factor.  If the power supply was of poor quality, which I would doubt in this case, it would have had a shorter amount of time to act upon the drives, as they were killed off far faster than would happen in real life.  These failures were as close to pure NAND wear-out by read-write cycles as is possible.  Other models of drives, and particularly those exposed to the harsher conditions that laptops often have to endure, may behave differently than these.

          There’s an interesting psychology that goes on with SSDs and their predetermined number of read/write cycles they can tolerate.  They have a lifespan that is relatively predictable from the start, while hard drives don’t.  It does not mean that hard drives last forever… just that their eventual failure isn’t as easy to predict.  The relative predictability of NAND wearout makes it seem more real, and that in turn makes people want to preserve the life they have more so than with hard drives, even though hard drives inevitably wear out too.

          Another thing I’ve seen people say is that with SSDs, backing up your data is more important than with hard drives.  I do not agree with that.  I would say it is as important… no more and no less.  Any data storage device can fail at any time, and every data storage device WILL fail at some point if left in service.  That includes the one you use for the backup, of course, but the more places you have your data, the less likely it is that disaster will strike at the same time on each of them.

          Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
          XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
          Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

    • #1923280

      I have a Crucial 250GB SSD from the days of Vista, Windows 7 & upgraded to Windows 10 and it still runs as good as new. I make sure the Optimization feature automatically to stay on schedule (in my case monthly by now) and use Crucial’s Storage Executive once a year during the month of August to ensure the “status” of the SSD to operate as it should.
      I have to admit I’m not as adept on how many of you can do with the set-ups you do as I’m a simple average user but that reminds me that the information in these posts are so valuable.

      HP EliteBook 8540w laptop Windows 10 Pro (x64)

    • #1923390

      SSD’s may outlive an HDD but they will slow down eventually if they are not cleaned up with a “Trim” program.  I think Windows 10 has Trim, maybe you could find one for Win 7.

      I put a Samsung 250 GB 860 EVO SSD in my old Sony laptop that I have Linux Mint installed on, and linux has FSTrim that I use about every two weeks or so.

      Windows 10 has automatic TRIM on, and you can set TRIM on Windows 7.

    • #1923393

      Ascaris: ” And now, after 5.5 years, the drive reports that it still has 68 percent of its rated service life remaining.

      This sounds like something useful to know. How does one interrogate an SSD to get that information?

      Use CrystalDiskInfo.

      1 user thanked author for this post.
    • #1924126

      A good backup system is a must I believe you have one from another post. I would not wory too much with a good backup in place, but as SSDs are cheap these days dedicating one to swap file usage is not unreasonable. If your computer needs and uses a swapfile having it on an SSD would be the way to go. As SSDs seem to die w/o as much warning as HDDs using a HDD as a data drive would be what I would do, unless you do data intensive operations.

      PS Just putting one together myself but a little stalled :[

      🍻

      Just because you don't know where you are going doesn't mean any road will get you there.
    • #1924749

      Use the SSD anywhere you want speed. OS, swap, video editing etc.
      To save space put static data on an HDD. Old videos, music etc.

      SSD Life article.

      cheers, Paul

    Viewing 8 reply threads
    Reply To: How to Keep my SSD Alive for Longer

    You can use BBCodes to format your content.
    Your account can't use all available BBCodes, they will be stripped before saving.

    Your information: