Design And Implementation Of Data Stream Management Prototype System

With the recent development in the field of information management and application, such as internet monitoring, Web application, sensor network, and financial analysis, a new form of data, the data stream, has come under increasing attention. Data stream features continuous arrival of data, with speed and large scale. A typical application requirement for it is how to produce results of data disposal continuously along with the continuous arrival of the data in question. The traditional database management system is a collection of relations relatively static, not a design for speedy and continuous data stream, and unable to give direct support for continuous queries. Therefore, how to develop a new type of database system in an effective manner so as to satisfy this new data requirement has become a hot subject in present studies.The present study is intended to design a universally applicable data stream management system, ComStream, with exploration on how to realize such a prototype system. Compared with the conventional DBMS, ComStream is able to support the continuous query in data stream. This paper in the first place discusses the overall design of ComStream before going on to describe the kernel of the data structure of the system. Due to the boundlessness of data stream and the need for continuous query, the traditional DBMS mode of query cannot be applied to ComStream directly. The paper discusses the query processing, taking the window as part of the standard semantics of ComStream, solving to a certain extent of the boundless storage needs. Operators like Select, Project, Join and Aggregation are realized. The Join and Aggregation query are two query operations based on windows. The traditional Database approach cannot be directly applied. The present paper comes up with some improvements making it applicable to windows.