数据流查询处理中一种共享的分级窗口维护策略  

Strategy of Maintaining Shared Multi-Level Windows in Query Processing of Data Streams

作　　者：袁锋[1] 宋宝燕[2] 武珊珊[1] 于亚新[1] 于戈[1] 

机构地区：[1]东北大学信息科学与工程学院 [2]辽宁大学信息科学与技术学院辽宁沈阳110036

出　　处：《小型微型计算机系统》2005年第12期2227-2232,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金(60473073)资助;辽宁省自然科学基金项目(20022027)资助;国家"八六三"高技术计划CIMS主题(2004AA1Z2060)资助;教育部优秀青年教师科研教育奖励计划资助.

摘　　要：由于数据流具有无界的特性,数据流系统中的查询多为带有窗口的查询,对带有窗口的查询,现有方法常由操作符直接维护窗口,但操作符的类型及排列方式可能会导致窗口难以维护,且冗余度较大.因此提出一种查询处理中的分级窗口维护策略,将窗口分为流窗口和操作符窗口,以流窗口为主并控制操作符窗口的维护,使查询中的窗口保持一致,解决了窗口维护问题,并且符合流查询语言的语义,各级窗口中的数据通过共享来解决内存消耗问题.Windows are widely used in query processing of data streams because streams are unlimited. In many systems, windows are maintained by operators which are main parts of query plans. But some operators can＇t maintain their windows properly because they affect each other for various reasons. Maintaining multi-level windows in query processing is a method to solve these problems. The two levels of windows are stream level and operator level. Stream windows are added to streams and maintained according to semantic of query languages. Stream windows control the maintenance of operator windows to make the latter process consistently. Sharing multi-level windows and data, System can also improve the utility of memory.

关 键 词：数据流 分级窗口查询处理 操作符窗口 流窗口 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]