Extended Approach for the 2-Phase Query Evaluation (2PE) Scheme

Abstract

Based on the previously proposed 2PE scheme, this paper proposes a new scheme called a 2-Phase Multiple Query Evaluation (2PME). The proposed scheme extends the 2PE scheme in a way that enables multiple multi- way join continuous queries to be more efficiently processed. It constructs a global execution plan that clusters a common sub-query walk (SQW) shared by more than two queries. Each cluster is executed only one time instead of being executed repeatedly for every related query. Ongoing changes in the rapid data streams in a data stream management system (DSMS) may be monitored using a number of pre-registered queries that report results continuously as new data elements of the data streams arrive (1,2). These queries are called continuous queries. In a previous study, we proposed a 2-phase query evaluation (2PE) scheme for a single multi-way join continuous query. Given a continuous query over multiple data streams, a preprocessing phase detects whether newly arrived stream tuples satisfy the query by identifying the maximum number of candidate tuples that could satisfy it. Subsequently, a evaluation phase identifies the stream tuples that satisfy the query among the candidate tuples specified in the previous phase. In this paper, the proposed 2PE scheme is extended as a way to process multiple continuous queries. It is called a 2-phase multiple query evaluation (2PME) scheme. Given a multi-way join continuous query, the 2PE scheme has defined a query walk, which is a specific order in which all the join predicates of the query are processed. The main issue of the 2PME scheme is how to efficiently share the common sub-query walks among the multiple multi-way join continuous queries, based on the matrix-based approach of 2PE. The proposed scheme shares common sub-query walks by constructing a global execution plan where the sub-query walks that are applicable to more than two continuous queries are clustered. Each cluster is