Facts figures, fun from managing 100000 HDDs

### Cloud Storage Reliability Across 100,000 Hard Drives At 00:00, Andy Klein (Principal Cloud Storage Analyst at Backblaze) delivers an empirical analysis of hard drive reliability based on data collected from operating over 100,000 hard disk drives (HDDs) in high-density cloud data centers.

### Fleet Telemetry and Annualized Failure Rates (AFR) At 11:01, Klein presents longitudinal failure statistics across more than 132,000 drives from major manufacturers, including Seagate, HGST, Western Digital, and Toshiba. He explains how Backblaze calculates Annualized Failure Rates (AFR) to benchmark drive reliability across multi-petabyte storage pods.

### Drive Bathtub Curves and Generational Differences At 22:01, the speaker analyzes failure curves. While infant mortality spikes occur in early operating months, drive survival stabilizes before mechanical wear sets in after year four. Klein highlights noticeable performance variations between drive generations, noting HGST's sustained longevity and Seagate's marked improvements in later 8TB and 12TB enterprise lines.

### Predictive S.M.A.R.T. Metrics for Imminent Drive Death At 33:03, Klein details S.M.A.R.T. telemetry. Backblaze's statistical models reveal that among dozens of recorded attributes, five specific metrics strongly predict drive failure: S.M.A.R.T. 5 (Reallocated Sectors), S.M.A.R.T. 187 (Reported Uncorrectable Errors), S.M.A.R.T. 188 (Command Timeout), S.M.A.R.T. 197 (Current Pending Sectors), and S.M.A.R.T. 198 (Offline Uncorrectable Sectors).

### Supply Chain Economics and Helium-Filled Drives Concluding between 44:01 and 55:00, Klein recounts operational history, including the 2011 Thailand floods that forced engineers to buy retail drives at consumer big-box stores ("drive farming"). He concludes by evaluating the thermal and energy advantages of modern sealed helium drives over traditional air-filled models.

(AI summary — Gemini Flash)