# SMART Drive Monitoring **Entity class:** Storage health reporting standard / technology ## Definition and history Self-Monitoring, Analysis and Reporting Technology, usually SMART, exposes drive-health attributes and warnings to hosts. [[articles/Intelligence Inside|Intelligence Inside]] places its lineage after IBM's 1992 Predictive Failure Analysis and the subsequent IntelliSafe initiative. A device can measure or classify its own condition locally while reporting selected results outward. A fleet operator can then build a separate model across many drives. ## Mechanism and limits The local threshold, host-visible attribute, and fleet prediction occupy different layers of the system. Their presence does not establish that each drive is retraining a machine-learning model internally. Longitudinal aggregation turns isolated maintenance signals into population-level estimates of failure risk. ## Relationships **Precursor:** [[wiki/Predictive Failure Analysis|Predictive Failure Analysis]]. **Interfaces with:** [[wiki/Telemetry|Telemetry]] and [[wiki/Classifier|Classifier]]. **Institutional layer:** [[wiki/Predictive Engines|Predictive Engines]]. **Article route:** [[articles/Intelligence Inside|Intelligence Inside]]. ## Sources [SMART project history and PFA documentation](https://smartlinux.sourceforge.net/smart); [[articles/Intelligence Inside|Intelligence Inside]].