大规模图计算系统研究进展  

Research Progress on Large-scale Graph Computing System

作　　者：李金忠 彭蕾[1,2] 刘欢 罗文浪 

机构地区：[1]井冈山大学电子与信息工程学院计算机科学与技术系,江西吉安343009 [2]网络与数据安全四川省重点实验室电子科技大学,成都610054 [3]同济大学电子与信息工程学院计算机科学与技术系,上海201804 [4]井冈山大学流域生态与地理环境监测国家测绘地理信息局重点实验室,江西吉安343009

出　　处：《小型微型计算机系统》2017年第10期2394-2400,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61762052)资助;网络与数据安全四川省重点实验室开放课题(NDSMS201602)资助;江西省自然科学基金项目(20171BAB202010;20161BAB202049;2011ZBAB202008)资助

摘　　要：随着大数据的快速发展,如Web图和在线社交网络等大图已经变得普遍,大规模图计算也越来越重要.为了有效处理如此大规模图,近年来已开发了许多大规模图计算系统.本文对大规模图计算系统进行综述.首先,对现有大规模图计算系统进行分类和评论,并简要阐述了一些大规模图计算系统.其次,对比分析了大规模分布式图计算系统中的三大派系:MapReduce派系、Pregel派系和GraphLab派系.最后,对大规模图计算系统未来的发展趋势进行了展望.With the rapid growth of Big Data, massive graphs, such as web graphs and online social networks, have become common, the large-scale graph computing is also more and more important. To efficiently process such large-scale graphs, many large-scale graph computing systems have been developed in recent years. A comprehensive review of the large-scale graph computing system is provided. Firstly, existing large-scale graph computing systems are classified and reviewed, and some large-scale graph computing systems are described briefly. Secondly ,MapReduce family ,Pregel family and GraphLab family are compared and analyzed in the largescale distributed graph computing systems. Finally, some promising development trends for future research of large-scale graph computing system are pointed out.

关 键 词：图计算 MAPREDUCE模型 BSP模型 GAS模型 大数据处理 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]