ISSUE 19.06.F • 2022-02-07

ADVERTISEMENT

The secure enterprise password manager 1Password is the world’s most-loved password manager, with top ratings from G2Crowd and Trustpilot, and has been named top password manager by leading media outlets including Wired, The New York Times, and CNET. 1Password makes it simple to create and share unique, strong passwords. With 1Password, you can securely share everything you need to work together, and give employees access to logins, documents, credit cards, and more, on all their devices. Try 1Password Business free for 14 days!

HARDWARE

By Ben Myers

With solid-state drives (SSDs), the SMART ante is raised because an SSD can fail catastrophically — CLUNK! — without warning and with no possibility of recovering data.

In my recent article Hard drives — still pretty S.M.A.R.T. (AskWoody 2021-12-27), I was hardly overwhelmed by the treatment of the S.M.A.R.T. (Self-Monitoring, Analysis, and Reporting Technology) data kept on traditional spinning hard drives by Microsoft Windows and the rest of the industry. But at least, if your computer started to hiccup, you could almost always look at the SMART data to find a possible cause.

The availability of useful SMART data for SSDs remains problematic. There is clearly no publicly available standard for the SMART data elements in SSD controllers, in contrast to an agreed-upon standard for hard drives. A table presented near the end of this article, compiled from sometimes sketchy SSD-manufacturer information, confirms this lack of a standard. Beyond the data defined in the hard-drive standard, SSD manufacturers can add whatever elements they want — and they have. Even though SSDs have become affordable commodity products, manufacturers remain very secretive about the algorithms and methods they use to keep drives in good operating condition, and so they make less information available to the public. When I recently contacted SK Hynix, now owner of Intel’s SSD business, for information about the SMART attributes for its SSDs, the response was that the company divulges the information only on its website.

You can read data from an SSD almost forever, and it will be extremely unlikely to fail. But repetitive writing of information to an SSD ultimately wears out the SSD’s flash memory chips. Manufacturers have compensated by implementing various schemes, some proprietary, to manage the wear. It would be nice to know how much life remains in an SSD. It would be oh so nice to know before an SSD goes CLUNK! Truth be told, early SMART drives did go CLUNK! and became unreadable paperweights. When the available number of spare blocks gets to zero in modern SSDs, they become read-only, so you can at least recover data.

Ever more sophisticated algorithms have pushed the projected life of a typical SSD to five years or more, sometimes with matching warranty, in a typical desktop or laptop system used in a typical way. Yes, I know. What’s typical?

Heavy use of an SSD can bring symptoms of early wear and even failure. I once replaced a nonresponsive and unreadable 128GB SSD in a MacBook Air with only 4GB of memory, just a few weeks after its warranty expired. Why? With too little system memory in this bargain MacBook, MacOS was forced to swap program segments and data repeatedly onto its proprietary SSD, wearing it out. CLUNK! — a total failure.

To examine the SMART information available from SSDs of different manufacturers, I used the freeware program Clear Disk Info (download from OlderGeeks.com). Figure 1 shows the full extent of the program’s display, which contains comprehensive information and descriptions of each datum.

Figure 1. Intel SSD showing signs of premature aging

The upper left corner of the Window lists the drives in the system and allows the selection of the device for which detailed information will be displayed. In this case, a heavily-used Intel 256GB SSD with 6,922 hours in operation is selected.

6,922 hours is less than 300 full days of use, a relatively short amount of time. From the image at left, you can see the prediction that the drive has 97% remaining lifetime. That doesn’t sound too bad by itself, but other details present warnings that indicate heavier wear. Many vendor apps that display only basic data will stop with the 97% number and not reveal underlying problems.

Figure 2 shows some of the details for the Intel SSD. Although failure is not predicted, one of the critical attributes shows that 10 sectors have been reallocated. That is a tiny number compared to the total number of sectors available but it is an indicator that problems may be emerging.

Figure 2. Details from Clear Disk Info

The lessons here are twofold. First, an SSD is not a single remedy for good system performance. More system memory will reduce SSD wear. A heavily used computer running Windows without enough memory will have a fate similar to that of the MacBook Air. Second, install the largest SSD you can afford, which will spread the wear of flash memory across its greater capacity.

Earlier SSDs hewed to one of two formats, both conforming to the existing SATA drive specification. One, the mSATA SSD, is a card-like format often attached to the motherboards inside small lightweight devices, such as Chromebooks and the precursors of the now-popular ultrabook category. (mSATA stands for Mini-Serial ATA, an SSD standard.)

Figure 3 shows perhaps the most popular form factor of SSD in use, approximately 2.5” wide by 4” long by 3/8” high. It is an exact physical replacement for a conventional rotating hard drive with a SATA interface and is easily used in both desktop and laptop PCs.

Figure 3. This SSD looks like a 2.5” laptop hard drive.

Figure 4 shows two stick-like SSDs that look identical except for their edge connectors. The Western Digital drive at the top of the photo, marked NVMe (Non-Volatile Memory Express), has a single notch that needs a PCI Express (PCIe) connection. The Micron drive at the bottom of the photo, clearly labeled 512GB SATA 6GB, has two notches in its edge connector and requires either a cable or an adapter board to match up with a SATA port, or the proper M.2 slot on the motherboard. Both SSDs attach to a computer with a single screw after the SSD is inserted into a slot.

Figure 4. Deceptively similar SSD sticks, one NVMe and the other M.2

Most of the major name-brand manufacturers of SSDs provide utility software that shows useful information, including SMART data, but only for their own branded products. This software typically runs as advertised with a retail-packaged SSD, but it may or may not provide useful information with an OEM version of an SSD. Figure 5 shows the information, or lack thereof, when Samsung Magician runs in a system with one of its OEM SSDs.

Figure 5. Samsung Magician is not always magical.

Software programs access the SMART data in NVMe SSDs with different data structures from those used for SATA SSDs. Your best bet, especially if you are dealing with numerous computers from different manufacturers, is to use Clear Disk Info to provide useful examples of SMART information from the various SSDs. To the best of my knowledge, it is the only free software available that provides you with a clear picture of the SMART data for any type of drive — as well as an interpretation of what the data means for you and your drive. There is also licensed software available which provides the same data plus other information unrelated to disks.

Following are several examples to give you some idea of what you can expect from looking at the SMART data in various brands of SSDs at different times in their expected lives. The first two examples of SMART data tell you something is not right about an SSD.

Figure 6 shows data from an older Samsung 850 EVO drive with 1TB of capacity. It has been used heavily, 17,213 hours, and its SMART data indicates 39% of its lifetime remains. This drive needs watching! Ultimately, I cloned it and replaced it, because its laptop behaved unpredictably.

Figure 6. Heavily used Samsung 850 2.5” SSD

Figure 7 shows data telling us that an OWC SSD is failing, with one percent of its lifetime remaining (orange bar). I would not trust any of the OWC data, because the rest of it is inconsistent with that one-percent value. But maybe OWC is off the hook here, because the SSD came from a failed MacBook Pro — and Apple is known for not caring about SMART. The OWC drive goes to the e-scrap bin, never to see the inside of a computer again.

Figure 7. OWC 2.5” SSD, close to failure

The SMART data for some drives, such as the Intel 1TB NVMe as shown in Figure 8, is very extensive.

Figure 8. SMART data for Intel NVMe SSD

A small SanDisk SSD has pretty extensive SMART data, as shown in Figure 9. But once it is used more heavily, the only reliable measure of health is its Percent Lifetime Remaining attribute. The attribute Grown Bad Blocks is not helpful in isolation because the SMART data does not tell us how many spare blocks the drive had originally.

Figure 9. SMART data for SanDisk 2.5” SSD

Less-popular SSD brands cannot be depended upon to provide comprehensive SMART data. Figures 10 and 11 illustrate the limited information fot Silicon Power (SPCC) and PNY 2.5” SSDs.

Figure 10. SMART data for Silicon Power 2.5” SSD

Figure 11. SMART data for PNY 2.5” SSD

Monitor SSD health yourself with some regularity, but not obsessively. You will also need to be able to interpret the SMART attributes of your drive(s), so obtain an SSD manufacturer’s SMART information, if available. If your system has a retail SSD installed, the manufacturer’s own utility will provide useful information. Clear Disk Info is convenient and easy to use. It is portable (does not require installation), so it can run from a Hirens USB flash drive if need be. How often you monitor depends on your usage pattern: how many hours a day you use your computer, and what sort of write-intensive tasks you do. How often? That’s your call.

SSD SMART data shows some useful derived numbers, such as percentage of drive life remaining. I like to see raw numbers, too — for example, the original number of spare blocks of data, the total number of spare blocks consumed, and the total number of blocks that make up the rated capacity of a drive. Looking at the ratio of original spares to currently available spares would give a good approximation of drive health, a figure derived by Clear Disk Info and, sometimes, by manufacturer’s SSD utilities.

The numbers shown in some SSD SMART data are downright puzzling. Just look at the data compiled in Figure 12 below, and see whether you can make sense of some of the information reported in the earlier Crucial, Intel, SanDisk, and Samsung examples. (If you use your browser’s zoom feature, you’ll be able to see the details better.)

Figure 12. Important SMART data for SSDs, per manufacturer data

What if the major SSD manufacturers got together, just as the original hard-drive manufacturers did when they agreed on a SMART spec for hard drives, and hammered out a common standard specification for the common SMART data pertaining to SSDs? That would be a major event.

Only recently did Microsoft begin to recognize the value of SMART, with a sole reference relegated to a TechNet article that rightfully extolled CHKDSK to check and repair structural integrity of the file system. Otherwise, there may have been only sparse mentions of SMART by Microsoft. The article How to Check Drive Health and SMART Status in Windows 10 (TenForums, 2021-06-20) shows five different ways to look at some SMART information, using Windows commands and programs, but the information presented to us by Windows alone contains few details. The TenForums article also contains a table of SMART information similar to the one in this article. Clear Disk Info provides more information, more quickly — with interpretation of results, not just raw data — compared to what is built into Windows.

What we really need to see is hard-drive and SSD monitoring built into Windows itself, with pop-up warnings when drives are not working properly. Then we would all get early warnings of flawed SSDs and hard drives, so that the dreaded BSODs and nonbooting systems could be avoided. These conditions are now even more difficult to troubleshoot, as Microsoft continues to add security features to Windows.

Active SMART would move the computer industry light years ahead in dealing with system reliability. But today, even with its highly secure Windows 11, Microsoft leaves us all on our own.

Join the conversation! Your questions, comments, and feedback about this topic are always welcome in our forums!

For over 25 years, Ben Myers has offered “cradle-to-grave” computer services for small businesses and individual computer owners — including building, upgrading, refurbishing, testing, repairing, and recycling of computer and network gear, primarily with Windows, some MacOSes, and Linux.

ADVERTISEMENT

You’re welcome to share! Do you know someone who would benefit from the information in this newsletter? Feel free to forward it to them. And encourage them to subscribe via our online signup form — it’s completely free!

Like what you see in the AskWoody FREE newsletter? Become a PLUS member! As a Plus member, you’ll receive the full newsletter, including all our great content about Windows, Microsoft, Office, 365, PCs, MS-DEFCON Alert notifications, useful and safe freeware, and Susan Bradley’s sought-after patch advice. Plus membership also allows continuous access to the complete archive of nearly two decades of Windows Secrets and AskWoody Newsletters. Naturally, Plus members have all the benefits of free membership, including access to the popular AskWoody forums. The cost? We’re supported by donations — choose any amount for a one-year membership. Every little bit helps.  Join AskWoody PLUS Today!