Semantic Load Shedding for Prioritized Continuous Queries over Data Streams

摘要

A data stream management system (DSMS) has to handle high-volume and bursty data streams with large number of continuous queries. When an input rate of any data stream exceeds the system capacity, the DSMS has to shed load by dropping some fraction of unprocessed data items. In this paper, we propose a new load shedding algorithm for continuous queries over data streams. Unlike previous algorithms assuming that all queries are equally important, we consider the priority of each query so that more important queries make more convincing outputs. As a result, the proposed algorithm can support differentiated quality of services by exploiting semantics inherent to applications. We also report the experiment results confirming the benefits of the proposed algorithm.