A Genetic Algorithm for Load Balancing in Parallel Query Evaluation for Deductive Relational Data Bases

Datalog is a query language for deductive databases. This language allows to evaluate a query incrementally on a network of processes. The tuples flow among them in parallel in order to compute the solution to the query. A major issue in this evaluation is the problem of assigning the processes to processors in a multiprocessors system. Not only load balancing is wanted but lowering the communication costs among the processors is crucial. We have tackled this problem by using a GA that works on a specific fitness function that allows to meet these goals. The techniques and results are presented here.