SSDStreamer: Specializing I/O Stack for Large-Scale Machine Learning

SSDStreamer: Specializing I/O Stack for Large-Scale Machine Learning

This article presents SSDStreamer, an SSD-based caching system for large-scale machine learning. By using DRAM as stream buffer, instead of an upper-level cache, SSDStreamer significantly outperforms state-of-the-art multilevel caching systems on Apache Spark, while requiring much less DRAM capacity...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE MICRO 2019-09, Vol.39 (5), p.73-81
Hauptverfasser: Bae, Jonghyun, Jang, Hakbeom, Gong, Jeonghun, Jin, Wenjing, Kim, Shine, Jang, Jaeyoung, Ham, Tae Jun, Jeong, Jinkyu, Lee, Jae W.
Format: Artikel
Sprache:eng
Schlagworte:
Artificial intelligence
Bandwidth
Caching
Data storage
Deep learning
Distributed Systems
Machine learning
Metadata
Nonvolatile memory
Prefetching
Random access memory
Solid state devices
Storage Management
Support vector machines
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This article presents SSDStreamer, an SSD-based caching system for large-scale machine learning. By using DRAM as stream buffer, instead of an upper-level cache, SSDStreamer significantly outperforms state-of-the-art multilevel caching systems on Apache Spark, while requiring much less DRAM capacity.
ISSN:0272-1732
1937-4143
DOI:10.1109/MM.2019.2930497