Priority-based Operator Scheduling Strategy in Data Stream System
Aimed at the problems of data stream query processing, a scheduling strategy based on operator priority is presented. In the paper, a PriOperator Model is designed which comprehensively takes the factors related to the operators and the system running state into consideration. Through simulating the relationship between the operator factors by means of Linear Regression, the priorities can be acquired by the linear function, and the operators are normally scheduled. For the purpose of dynamically modifying the operator priority, the artificial neural network learning (ANN) algorithm is also introduced. It gives a modification towards the operator priority according to the system performance. Therefore, the dynamical scheduling of operators can be realized. The applications of the strategy over the model reveal their effectiveness over the memory consumption and output latency.
priority operator scheduling data stream
LI Maozeng WANG Dan DU Dongming
College of Computer Science, Beijing University of Technology, Beijing 100022,China
国际会议
武汉
英文
332-337
2007-07-25（万方平台首次上网日期，不代表论文的发表时间）