Solving Stochastic Optimization in Distributed Databases Using Genetic Algorithms

Stochastic query optimization problem for multiple join is addressed. Two sites model of Drenick and Smith [2] is extended to four relations stored at four different sites. The model of three joins stored at two sites leads to a nonlinear programming problem, which has an analytical solution. The model with four sites leads to a special kind of nonlinear optimization problem (P). This problem can not be solved analytically. It is proved that problem (P) has at least one solution and two new methods are presented for solving the problem. An ad hoc constructive model and a new evolutionary technique is used for solving problem (P). Results obtained by the two considered optimization approaches are compared.