The Research Of Continuous Query Model Based On Network Datastream

The processing of network on the Internet datastream become more important application. But the request of real time and non-static storage make the querying and processing of network datastream a challenge in the data management. Many kinds of measurement have been used to satisfied this request. However, most of database manage systems being in use restrict the performance of datastream processing, and are lack of concentration on datastream features. Since traditional database queries focus on stored datasets, which are updated less or not frequent relatively, its shortage will be found when large in-flight and fluctuating datastreams come. So, heavily stream-oriented applications tend to use a DBMS largely as an offline storage system.In this paper, we describe an adaptive continuous query model based on network datastream, and its building and strcture. Concerning the fluctuation of network datastream, we prefer adaptivity of continuous query as the key point of our design, and introduce the eddy building block. It is a kind of fine-grained datastream operator, which can be placed between other query operators and cooperate with them. It can provide tuple level routing performance. To adapt to such an fine-grained method, we reorganize some related query operators to get better cooperation with eddy and better performance. The typical operators we improved are grouped filter for select operation and SteMs for join operation. Meanwhile, we augment the structure of tuple. We append it the head structure similar to the one of network data packet to improve its routing ability. Finally, a model of datastream continuous query is presented.In chapter 1, we introduce the background of this paper. In chapter 2, we introduce the basic structure of eddy, and analyse its working principle. Meanwhile, we illustrate its advantage in reordering operators and introduce its join algorism and routing policy. In chapter 3, we optimize some query operator, based on eddy. In chapter 4, we present the framework model of our system. In chapter 5, we design the wrapper program and system backend program. In chapter 6, we give an conclusion to this paper.