抄録

◎王　　岩・北川博之（筑波大）

With the increase of streaming information sources, stream processing has been an important research issue. In this paper, we introduce the designated event-based stream processing scheme in the stream processing engine. Different from the traditional stream processing scheme, the query results are supposed to be generated only when tuples come from designated streams specified by the user. After introducing the proposed scheme, we consider its implementation. We discuss some important implementation internals of the stream processing engine and show an efficient designated event-based execution scheme. We also consider multi-queries which is a common situation in the stream processing engine. The experiment shows the advantages of the proposed execution scheme that the throughput of the stream processing engine would be increased.