DISK Aug 24, 2026
A manufacturer-by-manufacturer HDD failure-rate comparison circulated via Blocks & Files (Aug 7). Vendor-average stats are interesting but noisy — what matters is your fleet. The admin lesson: watch real signals on your own clusters — storage disk show -health, storage disk show -broken, EMS disk error events, and Active IQ trend data; keep per-node spare policy honest (RAID-DP/RAID-TEC survives two failures, but a third during a long rebuild of large-capacity drives is the classic loss scenario); and replace failing disks proactively rather than waiting for the aggregate to degrade.