Multi-view Feature-based SSD Failure Prediction: What, When, and Why
summary
Proposes MVTRF multi-view multi-task random forest that concatenates per-log attributes with histogram and sequence features extracted from long-term and short-term telemetry, classifies failures into eight types and predicts remaining lifespan as time segments. Trains offline and runs online; evaluates on ~70 million telemetry records from 300,000 SSDs (Tencent and public datasets) and reports average improvements of 46.1% precision / 57.4% recall over prior schemes. Uses similar decision extraction to surface key tree rules for root-cause clues and keeps the pipeline interpretable.
AI-written summary · about
chapters (5)
related talks
14mFailureMiner: A Joint Key Decision Mining Scheme for Practical SSD Failure Prediction...
20mMSFRD: Mutation Similarity based SSD Failure Rating and Diagnosis for Complex...
27mMaking Disk Failure Predictions SMARTer!
31mFinding the Needle in the Haystack: Predicting Storage Device Failures in...
05mGenerating Realistic Wear Distributions for SSDs (Work-in-Progress Report)
25mDeepview: Virtual Disk Failure Diagnosis and Pattern Detection for Azure
18mPredictive Network: How AI Forecasts System Failures
40mPredictive Maintenance
