Repurpose Old SATA Hard Drives for Cold Storage: Windows SMART Tests, Checksums and Backup Rules

The idea is sound, with one condition that matters more than anything else in the setup. The drives are a convenient copy, not the archive itself. Sherback says so too. Everything on those disks is also covered by his separate 3-2-1 backups. Keep that in mind as you read the rest.

The setup: finished files, slow disks, no extra spending​

Sherback's reasoning is simple:

• The drives were already paid for. He bought them years ago, so reusing them cost him only the dock.
• Fast storage should hold active files. Moving old demos and clips off his main drives frees up the expensive storage for things he uses every day.
• Capacity and speed don't matter much here. Three terabytes is small by modern standards, but it's enough for finished files that never change.
• Old drives stay out of the landfill. Two working disks get a second job.

He also says prices for small drives have recently "spiked significantly." The source gives no dated pricing data, so treat that as his impression, not a measured trend. The money argument that does hold up is narrower: if you already own a working drive, reusing it costs very little. That doesn't mean old used drives always beat new or refurbished ones on value.

Section summary: This is a reuse story, not a buying guide. If the drive is already in your drawer, putting it to work is close to free.

Why hard drives beat SSDs on a shelf (with conditions)​

Sherback's main technical argument is how each type of drive behaves without power. A hard drive stores data magnetically on its platters, and that recording doesn't need power to persist. An SSD stores data as electrical charge in NAND flash cells, and that charge slowly leaks when the drive is unpowered. Heat speeds up the leak, and so does wear on the cells. Designs that pack more bits into each cell, such as QLC, leave less room for error.

He cites the JEDEC client-SSD retention requirement. It's real, but it needs context. The JESD218 table, reproduced by Kingston, lists the client class as "Active Use (power on): 40° C 8 hrs/day" with "Retention Use (power off): 30° C 1 year". The standard says the retention criteria apply to drives written to their endurance rating limits and then being subjected to the specified retention time.

That makes it a minimum for a worn-out drive, not a prediction for your lightly used one. A Western Digital white paper adds two useful points:

• There is an inverse relationship between PE cycles and data retention - as PE cycles increase, data retention diminishes. In plain terms, the more a cell has been written and erased, the less time it holds data.
• Most modern SSDs will have power-on background tasks which monitor how long it has been since a specific NAND block has been written and will actively refresh blocks that are degrading. An SSD in a drawer can't run that maintenance.

Temperature matters a lot. Curtiss-Wright, summarizing the same standard, notes that a client class SSD must maintain its data integrity at the defined BER for only 500 hours at 52°C (less than 21 days) or 96 hours at 66°C (only four days). AnandTech's well-known analysis called those scary scenarios unrealistic for most home users, because a drive is usually stored somewhere in the house in room temperature, which tends to be below 30°C. A hot garage or a car in summer is a different story.

Hard drives fail differently. Sherback points to the mechanical parts: lubricant and bearings age, and a drive that sat for years may not spin up cleanly. The magnetic data is rarely what fails first. His logic is that a mechanical problem shows up when you test the drive, while charge leaking from flash cells gives no warning. That's a reasonable preference. It is not proof that an unplugged hard drive will stay readable for a set number of years.

Section summary: For a shelf copy, hard drives avoid the charge leakage that limits unpowered SSDs. But "magnetic data lasts" isn't the same as "the drive will spin up in 2031."

Old drives fail more often, and Backblaze's data is more complicated than that​

Sherback admits the risk: older drives move closer to failure, and he cites Backblaze's Drive Stats as evidence. The real picture is less tidy.

In its 2025 "bathtub curve" revisit, Backblaze reported these annualized failure rate (AFR) peaks:

Backblaze's conclusion was that drives are getting better and lasting longer in its environment. The 2025 data still shows a mostly flat failure rate followed by a steep rise late in life, so age does matter eventually.

Backblaze also lists caveats that matter for home users:

• It's a data center. The drives run in controlled, consistent conditions, not in a USB dock on a desk.
• Bulk purchases skew results. A bad batch can fail all at once.
• Retiring healthy drives changes the curve. Backblaze sometimes pulls drives that are still working, which removes them from the statistics before they can fail.
• Nothing in the data covers storage without power. These are drives spinning all day, not disks left in a drawer for three years.

Section summary: Backblaze's numbers say age eventually catches up with drives. They can't tell you whether your particular 2TB drive from 2014 will survive.

How to set this up safely on Windows​

Sherback's advice is short: run a SMART extended self-test, do a full surface read, then reconnect the drives periodically to verify checksums. Here's a longer version. The tool suggestions are based on general Windows experience, not on his article.

1. Connect the drive and check its health. Windows reports only basic disk health. For a full SMART extended (long) self-test, use the drive maker's utility or a SMART tool. Some USB docks don't pass all SMART commands through, so if the tool can't see the drive's details, connect it directly to a SATA port for testing.
2. Read the whole surface. A full read finds weak sectors before you trust the drive with anything. Running chkdsk with the /r switch on Windows scans the volume for bad sectors. A vendor diagnostic full scan works too.
3. Fail it if it fails. Sherback's rule: a drive that doesn't pass these checks shouldn't be used for anything, cold storage included.
4. Copy the files and record checksums. In PowerShell, Get-FileHash produces SHA-256 hashes. Save the hash list somewhere other than the drive itself. If the list is stored on the drive, the same failure can damage both the files and the record you'd check them against.
5. Keep real backups elsewhere. Sherback calls a single old drive what it is: one copy on a single point of failure. Without checksums, a bad sector can silently corrupt a file for years before you notice.
6. Reconnect on a schedule. Plug the drives in now and then, let them spin up, compare the hashes, and replace any drive that starts throwing errors.

Checksums only detect damage. They don't repair it. If a hash doesn't match, you restore the file from your 3-2-1 backups. That's why those backups have to exist.

Section summary: Test the drive before use, record hashes off the drive, check them on a schedule, and keep the real backups somewhere else.

The verdict​

Sherback is right about the use case. Old SATA drives make a cheap, browsable shelf for finished files, and they free up fast storage for the work that needs it. His wording goes too far in a few places. "Reliable when unplugged" overstates it, and the JEDEC figure is a minimum for a worn drive, not a typical lifespan. His own setup gets the most important part right: the cold drives are for convenience, the backups protect the data, and losing one disk costs him some time, not any files.

If you have a couple of old drives and a dock, it's a weekend project worth doing. Test the drives first, record your checksums, and don't count the drawer copy as your backup.