Bayesian Methodologies For Big Spatial Data That Avoids Covariance Matrix Inversion