Dynamic Active Storage for High Performance I/O

Chao Chen, Yong Chen

Research output: Contribution to conferencePaper

Abstract

Many High-End Computing applications in critical areas of science and technology are becoming more and more data intensive. These applications transfer large amounts of data from storage nodes to compute nodes for processing, which is costly and bandwidth consuming and often dominates the applications run time. Active storage provides a promising solution for these applications by moving appropriate processing tasks from processing nodes to storage nodes. The prior research has achieved considerable progress and developed several active storage models. However, the existing studies have neglected that it is not optimal to offload every operation to storage nodes, especially when data dependence exists among operations. Such data dependence often exists in applications, such as the Geographic Information System and medical image processing. In such situation, it often wastes bandwidth for the conventional active storage architectures, because it is needed to transfer dependent data set
Original languageEnglish
StatePublished - Sep 2012

Fingerprint Dive into the research topics of 'Dynamic Active Storage for High Performance I/O'. Together they form a unique fingerprint.

Cite this