Global Query Optimization Based on Multistate Cost Models for a Dynamic Multidatabase System

Global query optimization in a multidatabase system (MDBS) is a challenging issue since some local optimization information such as local cost models may not be available at the global level due to local autonomy. It becomes even more difficult when dynamic environmental factors are taken into consideration. In our previous work, a qualitative approach was suggested to build so-called multistate cost models to capture the performance behavior of a dynamic multidatabase environment. It has been shown that a multistate cost model can give a good cost estimate for a query run in any contention state in the dynamic environment. In this paper, we present a technique to perform query optimization based on multistate cost models for a dynamic MDBS. Two relevant algorithms are proposed. The first one selects a set of representative system environmental states for generating an execution plan with multiple versions for a given query at compile time, while the second one efficiently determines the best version to invoke for the query at run time. Experiments demonstrate that the proposed technique is quite promising for performing global query optimization in a dynamic MDBS. Compared with related work on dynamic query optimization, our approach has an advantage of avoiding the high overhead for modifying or re-generating an execution plan for a query based on dynamic run-time information.