计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (12): 123-132.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

迭代式MapReduce研究进展

李金忠,汤鹏杰,夏洁武,谭云兰   

  1. 井冈山大学 计算机科学与技术系,江西 吉安 343009
  • 出版日期:2015-06-15 发布日期:2015-06-30

Advances in iterative MapReduce

LI Jinzhong, TANG Pengjie, XIA Jiewu, TAN Yunlan   

  1. Department of Computer Science and Technology, Jinggangshan University, Ji’an, Jiangxi 343009, China
  • Online:2015-06-15 Published:2015-06-30

摘要: 迭代计算普遍存在于大数据处理中,而传统的MapReduce不能显式地支持迭代计算。近几年,研究者扩展和改进原始MapReduce,已开发了若干迭代式MapReduce以更好地为大数据处理而支持迭代计算。对迭代式MapReduce编程框架进行综合评述,较详细地阐述了这些研究成果,给出了它们各自的基本思想,并分析了它们各自的特点、优势和不足,且对比了它们所采用的一些技术。对迭代式MapReduce未来的发展趋势进行了展望。

关键词: MapReduce, 迭代计算, 迭代式MapReduce, 并行编程模型, 大数据处理

Abstract: Iterative computations are pervasive among big data processing, but the traditional MapReduce cannot explicitly support iterative computation. In recent years, researchers have extended and improved the original MapReduce, and have developed a number of iterative MapReduce to better support iterative computation for big data processing. A comprehensive review of iterative MapReduce programming framework is provided. These research achievements are described in detail. Their basic ideas are given. Their characteristics, advantages and disadvantages are analyzed for each framework, and some technologies that have been adopted in these frameworks are compared. Some promising development trends for future research of iterative MapReduce are pointed out.

Key words: MapReduce, iterative computation, iterative MapReduce, parallel programming model, big data processing