一种数据流处理环境下的节点副本放置方法.pdf

第36 卷第7 期 电 子 与 信 息 学 报 Vol.36No.7 2014 年7 月 Journal of Electronics Information Technology Jul. 2014 一种数据流处理环境下的节点副本放置方法 丁维龙* 韩燕波 (北方工业大学云计算研究中心 北京 100144) 摘 要：物联网环境下的许多应用表现为传感数据的连续流式处理，且系统往往通过节点的副本技术保障可用性。 但是，运行时副本的备份和放置存在内存和带宽等资源开销，产生处理的延迟。该文给出一种方法，根据运行时的 资源消耗以贪心方式放置节点的副本，折中了系统的可用性和开销。实际系统的仿真实验表明，在相同的条件下， 该方法相比传统的随机放置，能为系统提供更稳定的可用性。 关键词：物联网；数据流；可用性保障；副本放置；贪心算法 中图分类号：TP301 文献标识码：A 文章编号：1009-5896(2014)07-1755-07 DOI: 10.3724/SP.J.1146.2013.01051 A Replica Placement Method during Data Stream Processing Ding Wei-long Han Yan-bo (Research Center for Cloud Computing, North China University of Technology, Beijing 100144, China) Abstract: Many applications of Internet of Things (IoT) are performed by the continuous stream processing of the senor data and nodes’ replicas are required to guarantee system availability. However, the replicas’ backup and placement often bring the processing delay at run-time due to the consumption of resources such as memory and bandwidth. In this paper, a method is proposed as greedy fashion by the resources cost to place nodes’ replicas, which could tradeoff between the availability and overheads of the system. Moreover, in a practical system, the extensive experiments show that the availability of the proposed method can be provided in a more stable manner than the traditional random placement under the same conditions. Key words: Internet of Things (IoT); Data stream; Availability guarantee; Replica placement; Greedy algorithm 1 引言 的数据，一旦处理节点出现故障，一方面影响甚至 物联网环境下的许多应用呈现数据的多源并发 中断下游的数据接收和处理，另一方面导致未能传 和汇聚处理的特征，为此近年来针对无边界、实时 递的数据迅速积累，影响上游节点的处理过程，甚