摘要
传统数据库系统用于需要持久稳固的数据存储和复杂查询的应用,然而近几年许多的应用证明这种数据模式和查询范例是不适合的,如传感器数据位置跟踪、工厂装配线管理、金融证券管理、Internet流量监控、Web使用日志、电话呼叫记录、和事务日志的在线分析等,在这些应用背景中信息按照数据值序列的形式自然产生,都需要及时地处理大规模的潜在快速的异步的数据流.本文回顾近来数据流管理系统领域的相关工作和流项目的研究情况;分析了连续查询处理的新需求和挑战;重点研究包括数据模式,系统结构,连续查询语言,调度方法,相关算法和查询评价等关键技术.
Traditional DBMSs are adapted to applications that need persistent stores and complex query, but a lot of applications emerged in recent years have testified that these data model and query paradigms are unsuitable, such as sensor location tracking, workshop assembly line management, negotiable securities, Interact flux monitoring, Web log analyzing, phone call recording, online transaction log analyzing, etc. The information in these applications has been generated according to value series form naturally. All of these application need to process potential hugevolume high - speed asynchronous data streams timely. The paper reviews the past work related data stream systems and ongoing projects in this area, explores topics in new requirements and challenges in continuous query processing, and focuses on the key technique for DSMS including data stream models, architecture of DSMS, continuous query language, scheduling strategy, related algorithm and query evaluating etc.
出处
《佳木斯大学学报(自然科学版)》
CAS
2005年第4期505-510,共6页
Journal of Jiamusi University:Natural Science Edition
关键词
数据流
数据流管理系统
连续查询
查询评价
data streams
DSMS
continuous query
scheduling strategy
query evaluating