面向数据密集型应用的细粒度内存管理方案

doi:10.13190/j.jbupt.2017.03.013

北京邮电大学学报 ›› 2017, Vol. 40 ›› Issue (3): 91-96,103.doi: 10.13190/j.jbupt.2017.03.013

面向数据密集型应用的细粒度内存管理方案

郝晓冉, 倪茂, 王力玉, 陈岚

中国科学院微电子研究所 EDA中心, 北京 100029

收稿日期:2016-10-10 出版日期:2017-06-28 发布日期:2017-05-25
作者简介:郝晓冉(1980-),女,助研,E-mail:haoxiaoran@ime.ac.cn;陈岚(1968-),女,教授,博士生导师.

Fine-Grained Memory Management Scheme for Data Intensive Applications

HAO Xiao-ran, NI Mao, WANG Li-yu, CHEN Lan

Institute of Microelectronics of Chinese Academy of Sciences, the EDA Center of Chinese Academy of Sciences, Beijing 100029, China

Received:2016-10-10 Online:2017-06-28 Published:2017-05-25

摘要/Abstract

摘要： 提出并实现了一种细粒度内存管理方案.该方案以用户数据对象为粒度对内存进行管理，当需要将部分内存数据换出以获得空闲内存空间时，不再像传统操作系统以页为单位换出，而是将多个冷数据对象换出到底层存储设备中，最大限度地保证了被换出的数据近期内不会再次被访问，从而有效减少了额外的I/O开销.测试结果表明，本方案与操作系统页交换机制相比，系统响应速度平均提高了37.5%.

关键词: 计算机系统结构, 页交换, 数据密集型应用, 虚拟内存管理, 物理内存管理

Abstract: For data intensive applications, a fine-Grained memory management scheme was proposed. The object-granularity data exchange between dynamic random access memory (DRAM) and swap device scheme was achieved. When DRAM is to be exhausted, the proposed scheme will swap data objects with low access frequency out to swap device, which reduces extra system I/O effectively. Compared to Linux-swap system, the proposed scheme improves system performance up to 37.5% in average.

Key words: computer system architecture, page swapping, data intensive application, virtual memory management, physical memory management

中图分类号:

TP302.1

郝晓冉, 倪茂, 王力玉, 陈岚. 面向数据密集型应用的细粒度内存管理方案[J]. 北京邮电大学学报, 2017, 40(3): 91-96,103.

HAO Xiao-ran, NI Mao, WANG Li-yu, CHEN Lan. Fine-Grained Memory Management Scheme for Data Intensive Applications[J]. JOURNAL OF BEIJING UNIVERSITY OF POSTS AND TELECOM, 2017, 40(3): 91-96,103.

参考文献

[1] Chen C L P, Zhang Chunyang. Data-intensive applications, challenges, techniques and technologies:A survey on Big Data[J]. Information Sciences, 2014, 275(11):314-347.
[2] Vibha B, Rahul J. Big data analysis:Issues and challenges[C]//EESCO 2015. Visakhapatnam:IEEE Press, 2015:1-6.
[3] Mauerer W. Professional linux kernel architecture[M]. Beijing:Posts & Telecom Press, 2010:821-824.
[4] Svwedlik Y. Facebook accelerates storage performance with flashcache update[EB/OL]. (2013-10-09)[2016-10-02]. http://www.datacenterdynamics.com/content-tracks/servers-storage/facebook-accelerates-storage-performance-with-flashcache-update/82690.fullarticle.
[5] Albrecht C, Merchant A, Stokely M, et al. Janus:Optimal flash provisioning for cloud storage workloads[C]//2013 USENIX Technical Conference. San Jose:USENIX Press, 2013:91-102.
[6] Armas C. MySpace replaces storage with solid-state drive technology in 150 standard load servers[EB/OL]. (2009-12-14)[2016-10-02]. https://www.infoq.com/news/2009/12/myspace-ssd.
[7] Ko S, Jun S, Ryu Y, Kwon C, et al. A new linux swap system for flash memory storage devices[C]//ICCSA 2008. Perugia:IEEE Press, 2008:151-156.
[8] Ko S, Jun S, Kim K, Ryu Y. Study on garbage collection schemes for flash-based linux swap system[C]//ASEA 2008. Sanya:IEEEPress, 2008:13-16.
[9] Xu Guangxia, Wang Manman, Liu Yanbing. Swap-aware garbage collection algorithm for NAND flash-based consumer electronics[J]. IEEE Transactions on Consumer Electronics, 2014, 60(1):60-65.
[10] Park Y, Bahn H. Management of virtual memory systems under high performance PCM-based swap devices[C]//COMPSAC 2015. Taichung:IEEE Press, 2015:764-772.
[11] Anirudh B, Vivek S P. SSDAlloc:hybrid SSD/RAM memory management made easy[C]//2011 Conference on Networked Systems Design and Implementation. Boston:ACM Press, 2011:211-224.
[12] Anirudh B, Vivek S P. Better flash access via shape-shifting virtual memory pages[C]//SIGOPS 2013. Farmington:ACM Press, 2013:1-14.
[13] Jian Xu, Steben S. NOVA:a log-structured file system for hybrid volatile/non-volatile main memories[C]//2016 USENIX Conference on File and Storage Technologies. Santa Clara:USENIX Press, 2016:323-338.
[14] Mendel R, John K O. The design and implementation of a log-structured file system[J]. ACM Transactions on Computer Systems, 1992, 10(1):26-52.

面向数据密集型应用的细粒度内存管理方案

Fine-Grained Memory Management Scheme for Data Intensive Applications

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 1

编辑推荐

Metrics

本文评价