Improving the accuracy, adaptability, and interpretability of SSD failure prediction models

Abstract

Flash-based solid state drives represent an important storage tier in today's hyperscale data centers. Although solid state drives (SSDs) are relatively reliable, data center operators are interested in predicting future drive failures to administer drive replacement, data migration, and drive acquisition strategies. We analyze telemetry data from over 30,000 SSDs running live applications in Google's datacenters over a span of six years, for predicting and explaining SSD failures using machine learning techniques. We propose the use of 1-class isolation forest and autoencoder-based anomaly detection techniques for predicting previously unseen SSD failure types with high accuracy. We show that ignoring the minority class for training can improve the performance by up to 9.5% and if adaptability to dynamic environments is required, by up to 13%. Furthermore, this paper proposes to utilize 1-class autoencoders to enable model interpretability. In particular, our autoencoder-based approach enables reasoning about the causes that lead to SSD failures. Common to all approaches, we deploy a set of powerful feature selection techniques that improve the model performance by up to 1.3× and reduce training times by up to 1.8×.

How does solid‐state drives cluster perform for distributed file systems: An empirical study

“…Among various storage mediums, the Solid-State Drive (SSD), by virtue of its high performance, has been identified as a mainstream substrate for storage systems. Unlike Hard Disk Drives (HDDs), SSDs are composed of semiconductor chips and thereby providing exceptional performance compared to HDDs 5 . On the other hand, as the cost of the SSD keeps dropping 6 , it is becoming more affordable to deploy SSDs in a cluster form in the storage systems.…”

“…Unlike Hard Disk Drives (HDDs), SSDs are composed of semiconductor chips and thereby providing exceptional performance compared to HDDs 5 . On the other hand, as the cost of the SSD keeps dropping, 6 it is becoming more affordable to deploy SSDs in a cluster form in the storage systems.…”

“…FIGURE5 Qualitative visualisation of the generalisability of the fitted regression model on the 2-node SSD cluster under three randomly selected settings.…”

Animation Character Detection Algorithm Based on Clustering and Cascaded SSD

“…SSD has good detection speed and accuracy, being one of the best target detection algorithms in industrial production. SSD algorithm takes into account the detection speed of the YOLO algorithm and the detection accuracy of the fast RCNN algorithm [11]. However, the effect of small target detection is general [12] since SSD uses Conv4_ 3 when a low-level feature is applied to small target detection.…”