Exploit SMART Attributes and NAND Flash Wear Characteristics for SSD Failure Prediction
Offered By: USENIX via YouTube
Course Description
Overview
Explore a conference talk that delves into a novel approach for predicting SSD failures in high-performance computing environments. Learn about the Aging-Aware Pseudo Twin Network (APTN), a method that combines SMART attributes and device-level NAND flash wear characteristics to forecast SSD failures more accurately. Discover how this innovative technique improves upon existing prediction schemes, achieving significant increases in F1-score and True Positive Rate (TPR). Gain insights into the importance of considering device-level wear characteristics alongside traditional SMART data for enhancing the reliability and accessibility of HPC storage systems.
Syllabus
USENIX ATC '24 - Exploit both SMART Attributes and NAND Flash Wear Characteristics to Effectively...
Taught by
USENIX
Related Courses
Introduction to Artificial IntelligenceStanford University via Udacity Natural Language Processing
Columbia University via Coursera Probabilistic Graphical Models 1: Representation
Stanford University via Coursera Computer Vision: The Fundamentals
University of California, Berkeley via Coursera Learning from Data (Introductory Machine Learning course)
California Institute of Technology via Independent