Abstract—In recent years, with the rapid advancement of
cloud technology, data centers have always been considered one
of cloud services most important aspect to evaluate, and its
reliability and availability have been the focus of every IT
engineer’s attention. However, service interruption is the most
important factor to consider for every data center, affecting the
user experience, or causing loss in a business. Therefore,
automated fault prevention and monitoring of data center will
effectively improve the reliability of cloud services. Predictive
maintenance differs from traditional maintenance process (i.e.
routine maintenance and corrective maintenance), it evaluates
the state by performing device condition monitoring, and
according to the state, it predicts when maintenance should be
performed.
This research focuses on hard drive failure prediction, with
big data analysis and machine learning technology, we have
developed a Preventive Monitoring System (PMS). Utilizing
Prognostics and Health Management (PHM) to identify the
failure mechanism, and combining Self-Monitoring, Analysis
and Reporting Technology (SMART) to identify early signs of
abnormalities before the device fail. Finally, we use random
forest algorithm to construct the predictive model. This
research aims to develop a predictive monitoring system to
provide device condition monitoring and fault diagnosis,
thereby identifying the device malfunction and resolving it as
soon as possible, keeping the system maintained at optimal
condition.
Index Terms—Big data, preventive maintenance, failure
prediction, hard drive, random forest.
The authors are with the Department of Industrial Engineering and
Management, Yuan Ze University, Taiwan (e-mail:
iecjsu@saturn.yzu.edu.tw, s1038908@saturn.yzu.edu.tw).
Cite: Chuan-Jun Su and Jorge A. Quan Yon, "Big Data Preventive Maintenance for Hard Disk Failure Detection," International Journal of Information and Education Technology vol. 8, no. 7, pp. 471-481, 2018.