[an error occurred while processing this directive]

ZTE Communications ›› 2013, Vol. 11 ›› Issue (4): 32-39.DOI: DOI:10.3939/j.issn.1673-5188.2013.04.005

• • 上一篇    下一篇

CPPL: A New Chunk-Based Proportional-Power Layout with Fast Recovery

Jiangling Yin, Junyao Zhang, and Jun Wang   

  1. Department of Electrical Engineering & Computer Science, University of Central Florida, Orlando, Florida 32816, USA
  • 收稿日期:2013-05-23 出版日期:2013-12-25 发布日期:2013-12-25
  • 作者简介:Jiangling Yin (jyin@eecs.ucf.edu) received his MS degree in software engineering from the University of Macau in 2011. He is working towards his PhD degree in computer engineering from the Electrical Engineering and Computer Science Department, University of Central Florida. His research focuses on energy-efficiency computing and file/storage systems.

    Junyao Zhang (junyao@eecs.ucf.edu) received his MS degree in software engineering from Jilin University in 2009. He is currently a Ph.D student in the Computer Science Department at University of Central Florida. His research interests include scalability, reliability and energy-efficiency issues in file/storage systems.

    Jun Wang (jwang@eecs.ucf.edu) received his PhD degree in computer science and engineering from University of Cincinnati in 2002.He is currently an associate professor with tenure in Department of Electrical Engineering and Computer Science, University of Central Florida. He is the recipient of National Science Foundation Early Career Award 2009 (news report) and Department of Energy Early Career Principal Investigator Award 2005. He is currently an associate editor of IEEE Transactions on Parallel and Distributed Systems. He has authored more than 60 publications in premier journals and leading HPC and systems conferences proceedings.
  • 基金资助:
    This work is supported in part by the US National Science Foundation Grant CCF-0811413, CNS-1115665, CCF-1337244 and National Science Foundation Early Career Award 0953946.

CPPL: A New Chunk-Based Proportional-Power Layout with Fast Recovery

Jiangling Yin, Junyao Zhang, and Jun Wang   

  1. Department of Electrical Engineering & Computer Science, University of Central Florida, Orlando, Florida 32816, USA
  • Received:2013-05-23 Online:2013-12-25 Published:2013-12-25
  • About author:Jiangling Yin (jyin@eecs.ucf.edu) received his MS degree in software engineering from the University of Macau in 2011. He is working towards his PhD degree in computer engineering from the Electrical Engineering and Computer Science Department, University of Central Florida. His research focuses on energy-efficiency computing and file/storage systems.

    Junyao Zhang (junyao@eecs.ucf.edu) received his MS degree in software engineering from Jilin University in 2009. He is currently a Ph.D student in the Computer Science Department at University of Central Florida. His research interests include scalability, reliability and energy-efficiency issues in file/storage systems.

    Jun Wang (jwang@eecs.ucf.edu) received his PhD degree in computer science and engineering from University of Cincinnati in 2002.He is currently an associate professor with tenure in Department of Electrical Engineering and Computer Science, University of Central Florida. He is the recipient of National Science Foundation Early Career Award 2009 (news report) and Department of Energy Early Career Principal Investigator Award 2005. He is currently an associate editor of IEEE Transactions on Parallel and Distributed Systems. He has authored more than 60 publications in premier journals and leading HPC and systems conferences proceedings.
  • Supported by:
    This work is supported in part by the US National Science Foundation Grant CCF-0811413, CNS-1115665, CCF-1337244 and National Science Foundation Early Career Award 0953946.

PDF

摘要： In recent years, the number and size of data centers and cloud storage systems has increased. These two corresponding trends are dramatically increasing energy consumption and disk failure in emerging facilities. This paper describes a new chunk-based proportional-power layout called CPPL to address the issues. Our basic idea is to leverage current proportional-power layouts by using declustering techniques. In this way, we can manage power at a much finer-grained level. CPPL includes a primary disk group and a large number of secondary disks. A primary disk group contains one copy of available datasets and is always active in order to respond to incoming requests. Other copies of data are placed on secondary disks in declusterd way for power-efficiency and parallel recovery at a finer-grained level. Through comprehensive theoretical proofs and experiments, we conclude that CPPL can save more power and achieve a higher recovery speed than current solutions.

关键词: power proportionality, parallelism recovery, declustering, layout

Abstract: In recent years, the number and size of data centers and cloud storage systems has increased. These two corresponding trends are dramatically increasing energy consumption and disk failure in emerging facilities. This paper describes a new chunk-based proportional-power layout called CPPL to address the issues. Our basic idea is to leverage current proportional-power layouts by using declustering techniques. In this way, we can manage power at a much finer-grained level. CPPL includes a primary disk group and a large number of secondary disks. A primary disk group contains one copy of available datasets and is always active in order to respond to incoming requests. Other copies of data are placed on secondary disks in declusterd way for power-efficiency and parallel recovery at a finer-grained level. Through comprehensive theoretical proofs and experiments, we conclude that CPPL can save more power and achieve a higher recovery speed than current solutions.

Key words: power proportionality, parallelism recovery, declustering, layout