首页 | 本学科首页   官方微博 | 高级检索  
     

发布/订阅系统中的缓存副本一致性研究
引用本文:王峰,李立新,曹景源,潘聪. 发布/订阅系统中的缓存副本一致性研究[J]. 计算机应用, 2016, 36(6): 1510-1514. DOI: 10.11772/j.issn.1001-9081.2016.06.1510
作者姓名:王峰  李立新  曹景源  潘聪
作者单位:信息工程大学, 郑州 450001
基金项目:国家863计划项目(2012AA012704)。
摘    要:针对发布/订阅系统中缓存副本一致性维护问题,首先,对原有基于谣言的一致性维护算法进行改进,提出一种基于轨迹标签的谣言一致性维护算法。该算法通过在更新消息报文中添加节点轨迹信息,避免向已更新节点发送冗余更新消息。其次,为提高系统消息传播的可靠性,设计一种基于轨迹标签的层次化反馈恢复机制,结合发布/订阅系统推拉传输模式,减少反馈消息数目,避免反馈爆炸。模拟实验结果证明,改进的一致性维护算法降低了一致性维护消息开销和时间开销,提高了系统的可靠性和可扩展性。

关 键 词:发布/订阅系统  一致性维护  谣言算法  轨迹标签  分层  推拉模式  
收稿时间:2015-11-10
修稿时间:2016-01-21

Research on replication consistency of cache in publish/subscribe systems
WANG Feng,LI Lixin,CAO Jingyuan,PAN Cong. Research on replication consistency of cache in publish/subscribe systems[J]. Journal of Computer Applications, 2016, 36(6): 1510-1514. DOI: 10.11772/j.issn.1001-9081.2016.06.1510
Authors:WANG Feng  LI Lixin  CAO Jingyuan  PAN Cong
Affiliation:Information Engineering University, Zhengzhou Henan 450001, China
Abstract:Aiming at the replication consistency maintenance problem of cache in publish/subscribe systems, firstly, a new algorithm based on trace label was proposed to improve the consistency maintenance algorithm based on gossip. The trajectory information of nodes was added to update information message, which avoided sending redundant update messages to updated nodes. Secondly, in order to improve the reliability of message propagation, a hierarchical feedback recovery mechanism based on trajectory label was proposed, which combined push/pull transmission mode of publish/subscribe systems, reduced the number of feedback messages, and avoided feedback implosion. The simulation experimental results show that the improved consistency maintenance algorithm can reduce the message cost and time cost of the consistency maintenance, and improve the system's reliability and scalability.
Keywords:publish/subscribe system  consistency maintenance  gossip algorithm  trace label  hierarchy  pull/push mode  
点击此处可从《计算机应用》浏览原始摘要信息
点击此处可从《计算机应用》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号