AI-based Proactive Storage Failure Management in Software-Defined Data Centres.

Abstract — Proactive failure management is essential to alleviate potential risks of service unavailability and downtime in Software-Defined Data Centres (SDDCs). Artificial Intelligence (AI) models enable proactive failure management by predicting and addressing potential failures before they actually occur. This paper proposes an AI-based Proactive Storage Failure Management (APSFM) solution for intelligent data centre management. The proposed solution includes a four-stage framework that employs AI models to predict failures efficiently. The study uses Random Forest and Artificial Neural Network models as examples to predict disk failures by employing Self-Monitoring, Analysis, and Reporting Technology (SMART) attributes. The experimental results have shown that both models can achieve high prediction performance.