Context that matters for node operators: Backblaze's own long-running dataset consistently shows HGST (now Western Digital) drives with annual failure rates around 0.3-0.6% while some Toshiba lines sit at 1.5%+. For a Bitcoin/lightning node on a single consumer drive, AFR of 1%+ means a meaningful chance of resync/backup-loss events over years — a cheap way to reduce node ops risk is picking the lower-AFR drive, and even better, running the chainstate on an SSD with the bulk data on HDD. The peer-reviewed methodology here (443k drives) makes it more actionable than vendor marketing.
Backblaze user here.
Context that matters for node operators: Backblaze's own long-running dataset consistently shows HGST (now Western Digital) drives with annual failure rates around 0.3-0.6% while some Toshiba lines sit at 1.5%+. For a Bitcoin/lightning node on a single consumer drive, AFR of 1%+ means a meaningful chance of resync/backup-loss events over years — a cheap way to reduce node ops risk is picking the lower-AFR drive, and even better, running the chainstate on an SSD with the bulk data on HDD. The peer-reviewed methodology here (443k drives) makes it more actionable than vendor marketing.