On transformation of query scheduling strategies in distributed and heterogeneous database systems

This work considers a problem of optimal query processing in heterogeneous and distributed database systems. A global query sub- mitted at a local site is decomposed into a number of queries processed at the remote sites. The partial results returned by the queries are in- tegrated at a local site. The paper addresses a problem of an optimal scheduling of queries that minimizes time spend on data integration of the partial results into the final answer. A global data model defined in this work provides a unified view of the heterogeneous data structures located at the remote sites and a system of operations is defined to ex- press the complex data integration procedures. This work shows that the transformations of an entirely simultaneous query processing strate- gies into a hybrid (simultaneous/sequential) strategy may in some cases lead to significantly faster data integration. We show how to detect such cases, what conditions must be satisfied to transform the schedules, and how to transform the schedules into the more efficient ones.