北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2014, Vol. 37 ›› Issue (1): 52-56.doi: 10.13190/j.jbupt.2014.01.012

• 论文 • 上一篇    下一篇

具有健康节点协作的高效多节点修复方案

谢显中, 王柳苏, 黄倩, 马彬   

  1. 重庆邮电大学 宽带接入网络研究所, 重庆 400065
  • 收稿日期:2013-05-10 出版日期:2014-02-28 发布日期:2014-01-07
  • 作者简介:谢显中(1966—),男,教授,E-mail:xiexzh@cqupt.edu.cn.
  • 基金资助:

    国家自然科学基金项目(61271259,60872037);重庆市自然科学基金项目(CTSC2011jjA40006,CSTC2010BB2415);重庆市教委科学技术研究项目(KJ120501,KJ110530)

Efficient Multi-Node Regenerating Program with Healthy Nodes Collaboration in Distributed Storage Systems

XIE Xian-zhong, WANG Liu-su, HUANG Qian, MA Bin   

  1. Institute of Broadband Access Networks, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • Received:2013-05-10 Online:2014-02-28 Published:2014-01-07

摘要:

针对分布式存储中多节点再生修复的修复带宽和可靠性问题,提出了一种具有健康节点协作的多节点修复方案. 该方案在修复模型上做出了相应改进,通过健康节点间的协作把上述问题做了折中. 给出了具体的多节点修复过程和相应再生码的构造,用信息流图求得最大流最小割集来说明其最小边界. 结果表明,在确保修复带宽最低的条件下使修复时间同步,修复过程更简便,最重要的是所需传输信道更少,保证了修复的可靠性.

关键词: 分布式存储, 多节点修复, 修复带宽, 修复可靠性, 健康节点的协作

Abstract:

In distributed storage systems, an efficient multi-node regenerating program with healthy nodes collaboration was proposed to make a trand-off between the repair bandwidth and reliability for multi-node regeneration. Specific multi-node repair process and the construction of regenerating code were introduced, and the max-flow-min-cut of information flow graph was used to illustrate its lower bound. It is shown that the program ensures the repair bandwidth maintain to minimum during the repair process, and makes the repair time synchronization and repair process easier. The most important is less that of desired transmission channels when ensuring reliability of regeneration.

Key words: distributed storage, multi-node repair, repair bandwidth, repair reliability, healthy nodes collaboration

中图分类号: