Query Plans - Their Cost and Generation in a Distributed Heterogeneous Environment

As with other distributed database systems, it is important in an Active Object Oriented Knowledge Base Management System (AOOKBMS) to be able to execute distributed queries efficiently and cost effectively. In this paper, we show how this task is being tackled in an ongoing project at the University of Florida dealing with Virtual Enterprises (VE). This project uses an AOOKBMS to not only describe a common view of the VE’s data and processing capabilities, but to also support distributed query optimization and processing. Through the use of wrappers this system allows the distributed execution of a query defined in an Object Query Language. It can be executed on heterogeneous systems that may include Object Oriented and Relational databases as well as non-database oriented systems. The results of this distributed processing is then combined and presented in a single uniform result, thus supporting the concept of a truly Virtual Enterprise. In this paper we cover how query optimization can be achieved using the AOOKBMS to store and manipulate all aspects of the query optimization and distributed processing tasks. In this paper we show that the AOOKBMS is a suitable tool for this task. To do this we present a cost function for any query tree defined in the AOOKBMS as well as an algorithm for turning a centralized query tree into a distributed query plan. Both this algorithm as well as the execution control mechanism are then presented using the implementation programming language developed for the