Disk Failure Prediction for Software-Defined Data Centre (SDDC)
Yongqing Zhu, Paul Horng Jyh Wu, Fang Cherry Liu, Renuga Kanagavelu · 2021 IEEE Intl Conf on Dependable, Autonomic and Secure Computing, Intl Conf on Pervasive Intelligence and Computing, Intl Conf on Cloud and Big Data Computing, Intl Conf on Cyber Science and Technology Congress (DASC/PiCom/CBDCom/CyberSciTech) · 2021
Efficient disk failure prediction is required to prevent data loss and service degradation in Software-Defined Data Centre (SDDC). Self-Monitoring, Analysis, and Reporting Technology (SMART) attributes have been widely studied for monitoring disk health. While the traditional threshold-based warning methods only achieve low failure detection, machine learning methods have been applied to SDDC for disk failure prediction. This paper presents a Multi-Layer Perceptron (MLP) model to predict hard disk failure based on SMART attributes. The MLP model includes four layers with fully-connected neurons. We use the SMART data collected from enterprise data centre and adjust the model with different parameters. Performance evaluation results have shown that the proposed MLP can achieve high accuracy on disk failure predication.