Explore a practical approach to predicting storage device failures in data centers through this 31-minute conference talk from SREcon23 Asia/Pacific. Delve into the challenges faced by Site Reliability Engineers in managing and monitoring vast numbers of storage devices, and learn about a multi-phase proactive sampling-based system designed to address these issues. Witness a live demonstration of the system implemented in a multi-tiered cloud storage pool, and gain insights into innovative techniques for improving accuracy, performance, and cost-effectiveness in failure prediction. Discover how this research can be applied to solve real-world challenges in data center management and storage device reliability.