美国东北大学糜宁芳副教授学术报告

发布日期:2019-06-27 浏览次数: [字体: ]

报告题目:Breaking Block with Key-Value: A New NVMe Key-Value Flash Storage Framework for HPC Applications

报告人: 糜宁芳(Ningfang Mi)博士

                  (美国东北大学副教授)

报告时间2019. 7. 2 (周二)下午3:30

报告地点:明理楼109

报告摘要

       Modern data centers are shifting from hard disk drives (HDDs) to solid state drives (SSDs) for data storage because of SSD's performance benefits. However, even with SSDs, I/O performance continues to be the bottleneck in high-performance computing (HPC) systems. One of the main reasons lies in the fact that conventional SSDs are block-based devices, which require conversions of application data into blocks to store data. Such intermediate data conversions are time-consuming and impede utilizing the full performance of NVMe SSDs for parallel HPC applications.  In this talk, we present a novel Key-Value based Storage infrastructure for Parallel computing (KV-SiPC) - a new method for multi-threaded OpenMP applications to use NVMe Key-Value SSDs. KV-SiPC simplifies the application data management by removing intermediate layers from the I/O stack of operating system (OS) kernel. We implement our KV-SiPC infrastructure in a real HPC system by extending various processing layers (e.g., program, OS, and device layers) and evaluate the performance of KV-SiPC by comparing block SSDs with the traditional I/O stack. The experimental results show that KV-SiPC can significantly speed up HPC applications by 10x - 120x when compared to that with traditional NVMe SSDs.

 

个人简介

      Dr. Ningfang Mi is an Associate Professor in the Department of Electrical and Computer Engineering at Northeastern University. She received her Ph.D. from the College of William and Mary, Department of Computer Science in August 2009. She received her M.S in Computer Science from the University of Texas at Dallas in 2004 and her B.S. in Computer Science from Nanjing University in 2000. She was a recipient of 2015 NSF CAREER Award, 2014 AFOSR Young Investigator Research Program (YIP) Award, and 2010 IBM Faculty Award. Dr. Mi has a strong background in resource allocation and scheduling, storage systems, performance evaluation, capacity planning, and workload analysis.