Imagine you are working with a massive dataset and need to implement a linear regression model using Spark. Describe the steps you would take to ensure that the model training is optimized for performance, including data preprocessing, choosing the right Spark configuration, and utilizing Spark's built-in functions for efficient computation.