data security · Storage

Used Drives: Reading SMART, Hours and Wear

· 2 min read · Server Depot

Used Drives: Reading SMART, Hours and Wear

Used enterprise drives are either the best bargain in storage or a false economy, and the difference is legible in about ninety seconds per drive. Enterprise disks log their whole lives; buying them well is mostly a matter of reading before trusting — and having a policy about what used storage is for.

Spinning drives: the fields that predict

Power-on hours gives context, not verdict: enterprise drives are engineered for continuous duty, and the 20–45k-hour band is normal mid-life. Treat hours as a pricing input. The predictive fields are the sector counters: reallocated sectors (grown defects the drive has mapped out), pending sectors (unreadable, awaiting verdict), and uncorrectable errors. The practical policy is strict: zeros across all three or the drive is a return. Large fleet studies agree with the strictness — drives with their first reallocations fail at multiples of the clean population's rate. Also glance at load cycle counts (aggressive parking histories) and spin-up behavior in the log.

SSDs: wear is arithmetic

Flash publishes its own consumption: percentage used / wear leveling (the honest odometer), total bytes written against the model's TBW/DWPD rating, and available spare. A power-loss-protected enterprise SSD at 15% wear after 30k hours has, functionally, most of its life ahead; the same model at 85% is priced-per-remaining-write territory. Two hard rules: available-spare warnings mean no purchase at any price, and datacenter SSDs without power-loss protection do not exist — if the model lacks PLP, it was never a datacenter SSD, whatever the listing says.

Context checks beyond the drive

Match firmware within an array (mixed revisions cause the weird 2 a.m. bugs), confirm sector format compatibility with your controller (512e vs 4Kn, and unformat 520-byte SAN pulls before assuming a bargain), and prefer drives whose model appears on your backplane's known-good list — expander quirks are real on older enclosures.

The policy that makes used drives safe

Used drives belong in arrays designed for members to die: RAID6/RAIDZ2 with a hot spare, mirrors with monitored counters, backup tiers whose loss costs a rebuild rather than data. They do not belong as single copies of anything irreplaceable — but then, no drive does. Buy tested drives with published SMART from sellers who grade honestly, scrub monthly so latent errors surface on your schedule, and replace at the first reallocation rather than the fifth. Run that policy and used enterprise storage delivers its real promise: capacity at a fraction of retail, with failures that arrive as maintenance instead of surprise.